)]}'
{"doc/source/devref/openvswitch_agent.rst":[{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"9f825655d2b8121d74049ac5141c524026dd04bc","unresolved":false,"context_lines":[{"line_number":298,"context_line":"especially if the end result would look a lot like other maturing"},{"line_number":299,"context_line":"alternatives."},{"line_number":300,"context_line":""},{"line_number":301,"context_line":"Implementation VLAN Interfaces (Option A)"},{"line_number":302,"context_line":"~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~"},{"line_number":303,"context_line":""},{"line_number":304,"context_line":"Trunk port creation"}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_10594b8e","line":301,"range":{"start_line":301,"start_character":0,"end_line":301,"end_character":41},"updated":"2016-05-23 17:30:43.000000000","message":"we should proably mention that option a cannot be\nused with vhost-user since we do not have any tap devices.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"808e23340e1d9899241b574767337836effcd200","unresolved":false,"context_lines":[{"line_number":298,"context_line":"especially if the end result would look a lot like other maturing"},{"line_number":299,"context_line":"alternatives."},{"line_number":300,"context_line":""},{"line_number":301,"context_line":"Implementation VLAN Interfaces (Option A)"},{"line_number":302,"context_line":"~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~"},{"line_number":303,"context_line":""},{"line_number":304,"context_line":"Trunk port creation"}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_1241163a","line":301,"range":{"start_line":301,"start_character":0,"end_line":301,"end_character":41},"in_reply_to":"bab6814e_10594b8e","updated":"2016-05-25 14:58:40.000000000","message":"isn\u0027t that mentioned already in the previous paragraph?","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"6c9137baee05f6146dd33215b6c656aa3ccc75bf","unresolved":false,"context_lines":[{"line_number":298,"context_line":"especially if the end result would look a lot like other maturing"},{"line_number":299,"context_line":"alternatives."},{"line_number":300,"context_line":""},{"line_number":301,"context_line":"Implementation VLAN Interfaces (Option A)"},{"line_number":302,"context_line":"~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~"},{"line_number":303,"context_line":""},{"line_number":304,"context_line":"Trunk port creation"}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_9f0003ad","line":301,"range":{"start_line":301,"start_character":0,"end_line":301,"end_character":41},"in_reply_to":"bab6814e_1241163a","updated":"2016-05-26 12:00:49.000000000","message":"I think it\u0027s good to write about these limitations as explicit as possible.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"0834b467007206792fa912de8be12b376591c080","unresolved":false,"context_lines":[{"line_number":311,"context_line":"will store the port ID of the trunk. The OVS agent detects that a new interface"},{"line_number":312,"context_line":"has been plugged. It gets the details of the new port after sending an RPC"},{"line_number":313,"context_line":"message (get_devices_details_list_and_failed_devices) to the Neutron server."},{"line_number":314,"context_line":"The details returned will have a new boolean field \"is_trunk\", to specify that"},{"line_number":315,"context_line":"the port is a trunk port and a field \"sub_ports\" to return all the details"},{"line_number":316,"context_line":"of the subports associated to this trunk. If the trunk port is not associated"},{"line_number":317,"context_line":"to any subport, the L2 agent will install a drop flow to drop all the traffic,"}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_1806d485","line":314,"updated":"2016-05-20 23:15:31.000000000","message":"This actually remain to be seen, because the RPC interface is still TBD. Besides, get_devices_details_list_and_failed_devices is something specific to ML2 and we don\u0027t want to pollute that with trunk details. We may want to come up with a mechanism like the one defined for QoS.\n\nFor the sake of this discussion, I\u0027d suggest you shy away on how you get to know that a port is a part of a trunk and limit the discussion on how the trunk creation happens without specifying how you get hold of the trunk information.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"808e23340e1d9899241b574767337836effcd200","unresolved":false,"context_lines":[{"line_number":311,"context_line":"will store the port ID of the trunk. The OVS agent detects that a new interface"},{"line_number":312,"context_line":"has been plugged. It gets the details of the new port after sending an RPC"},{"line_number":313,"context_line":"message (get_devices_details_list_and_failed_devices) to the Neutron server."},{"line_number":314,"context_line":"The details returned will have a new boolean field \"is_trunk\", to specify that"},{"line_number":315,"context_line":"the port is a trunk port and a field \"sub_ports\" to return all the details"},{"line_number":316,"context_line":"of the subports associated to this trunk. If the trunk port is not associated"},{"line_number":317,"context_line":"to any subport, the L2 agent will install a drop flow to drop all the traffic,"}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_7286a266","line":314,"in_reply_to":"bab6814e_1806d485","updated":"2016-05-25 14:58:40.000000000","message":"OK, we can discuss the RPC inteface later","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"135ed97d6ed44e85cfdffd8e3826ba0bdb50898e","unresolved":false,"context_lines":[{"line_number":311,"context_line":"will store the port ID of the trunk. The OVS agent detects that a new interface"},{"line_number":312,"context_line":"has been plugged. It gets the details of the new port after sending an RPC"},{"line_number":313,"context_line":"message (get_devices_details_list_and_failed_devices) to the Neutron server."},{"line_number":314,"context_line":"The details returned will have a new boolean field \"is_trunk\", to specify that"},{"line_number":315,"context_line":"the port is a trunk port and a field \"sub_ports\" to return all the details"},{"line_number":316,"context_line":"of the subports associated to this trunk. If the trunk port is not associated"},{"line_number":317,"context_line":"to any subport, the L2 agent will install a drop flow to drop all the traffic,"},{"line_number":318,"context_line":"since no traffic is supposed to flow when no subport is defined."},{"line_number":319,"context_line":"After that it will send a message notifying that the trunk port is up."}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_705153d8","line":316,"range":{"start_line":314,"start_character":0,"end_line":316,"end_character":41},"updated":"2016-05-19 13:31:16.000000000","message":"This is a bit redundant. We could omit the \"is_trunk\" field and return the \"subports\" field always. If it has zero length, then the port is not a trunk port. If it has one or more subports, it is a trunk port.\n\nBy the way in option (A) it doesn\u0027t matter (for the agent) if the port being processed is trunk or not, so we never need to check.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"d18809a5438e92881d1352f1211a539fed0f16c4","unresolved":false,"context_lines":[{"line_number":311,"context_line":"will store the port ID of the trunk. The OVS agent detects that a new interface"},{"line_number":312,"context_line":"has been plugged. It gets the details of the new port after sending an RPC"},{"line_number":313,"context_line":"message (get_devices_details_list_and_failed_devices) to the Neutron server."},{"line_number":314,"context_line":"The details returned will have a new boolean field \"is_trunk\", to specify that"},{"line_number":315,"context_line":"the port is a trunk port and a field \"sub_ports\" to return all the details"},{"line_number":316,"context_line":"of the subports associated to this trunk. If the trunk port is not associated"},{"line_number":317,"context_line":"to any subport, the L2 agent will install a drop flow to drop all the traffic,"},{"line_number":318,"context_line":"since no traffic is supposed to flow when no subport is defined."},{"line_number":319,"context_line":"After that it will send a message notifying that the trunk port is up."}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_b2767fa1","line":316,"range":{"start_line":314,"start_character":0,"end_line":316,"end_character":41},"in_reply_to":"bab6814e_2d0a3c26","updated":"2016-05-23 18:42:40.000000000","message":"I think in Option C the VIF for a regular port could just plug directly into br-int. If that were the case, regular ports can\u0027t be upgraded to trunk ports w/out downtime, but a migration of legacy ports to the new trunk-wiring wouldn\u0027t be required. I don\u0027t think we\u0027ll be able to get away with requiring all active ports to be migrated to the new trunk-wiring in order to use vlan-aware VMs, so we might need trunk ports to be wired differently than regular ports and have them coexist.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"366b54d9ab2de36836de9ac72dd794227f4bffca","unresolved":false,"context_lines":[{"line_number":311,"context_line":"will store the port ID of the trunk. The OVS agent detects that a new interface"},{"line_number":312,"context_line":"has been plugged. It gets the details of the new port after sending an RPC"},{"line_number":313,"context_line":"message (get_devices_details_list_and_failed_devices) to the Neutron server."},{"line_number":314,"context_line":"The details returned will have a new boolean field \"is_trunk\", to specify that"},{"line_number":315,"context_line":"the port is a trunk port and a field \"sub_ports\" to return all the details"},{"line_number":316,"context_line":"of the subports associated to this trunk. If the trunk port is not associated"},{"line_number":317,"context_line":"to any subport, the L2 agent will install a drop flow to drop all the traffic,"},{"line_number":318,"context_line":"since no traffic is supposed to flow when no subport is defined."},{"line_number":319,"context_line":"After that it will send a message notifying that the trunk port is up."}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_bf137c65","line":316,"range":{"start_line":314,"start_character":0,"end_line":316,"end_character":41},"in_reply_to":"bab6814e_705153d8","updated":"2016-05-19 16:32:58.000000000","message":"I think it will be possible to have a trunk port without any subport created on it (yet). In that case it will essentially act like a regular port today. But for option A, I don\u0027t think the distinction matters because a trunk port and a regular port will look exactly the same. For option C, just checking the subport list might not be distinctive enough, because a trunk port might be wired differently than a regular port.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"9f825655d2b8121d74049ac5141c524026dd04bc","unresolved":false,"context_lines":[{"line_number":311,"context_line":"will store the port ID of the trunk. The OVS agent detects that a new interface"},{"line_number":312,"context_line":"has been plugged. It gets the details of the new port after sending an RPC"},{"line_number":313,"context_line":"message (get_devices_details_list_and_failed_devices) to the Neutron server."},{"line_number":314,"context_line":"The details returned will have a new boolean field \"is_trunk\", to specify that"},{"line_number":315,"context_line":"the port is a trunk port and a field \"sub_ports\" to return all the details"},{"line_number":316,"context_line":"of the subports associated to this trunk. If the trunk port is not associated"},{"line_number":317,"context_line":"to any subport, the L2 agent will install a drop flow to drop all the traffic,"},{"line_number":318,"context_line":"since no traffic is supposed to flow when no subport is defined."},{"line_number":319,"context_line":"After that it will send a message notifying that the trunk port is up."}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_2d0a3c26","line":316,"range":{"start_line":314,"start_character":0,"end_line":316,"end_character":41},"in_reply_to":"bab6814e_bf137c65","updated":"2016-05-23 17:30:43.000000000","message":"how would the trunk port be wired differently for option c then a regular port?","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"b4da8f06477279f40c186847243451bd4f6c48da","unresolved":false,"context_lines":[{"line_number":313,"context_line":"message (get_devices_details_list_and_failed_devices) to the Neutron server."},{"line_number":314,"context_line":"The details returned will have a new boolean field \"is_trunk\", to specify that"},{"line_number":315,"context_line":"the port is a trunk port and a field \"sub_ports\" to return all the details"},{"line_number":316,"context_line":"of the subports associated to this trunk. If the trunk port is not associated"},{"line_number":317,"context_line":"to any subport, the L2 agent will install a drop flow to drop all the traffic,"},{"line_number":318,"context_line":"since no traffic is supposed to flow when no subport is defined."},{"line_number":319,"context_line":"After that it will send a message notifying that the trunk port is up."},{"line_number":320,"context_line":"Neutron will send back to Nova an event to signal that the wiring was successful."},{"line_number":321,"context_line":""}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_9eececc8","line":318,"range":{"start_line":316,"start_character":42,"end_line":318,"end_character":64},"updated":"2016-05-18 22:43:30.000000000","message":"Is this statement correct? I thought that if a trunk port isn\u0027t associated to any subport, then the trunk port would only pass untagged traffic that is destined for the trunk port\u0027s network. For example, I follow these steps:\n1) create a neutron port on network N - port P\n2) create a trunk port using port P\u0027s uuid (don\u0027t assign any subports)\n3) `nova boot` by passing it port P\u0027s uuid\nAt that point, my VM should have connectivity to network N via the untagged interface (eth0). I.e. a trunk port with no subports will basically act like a regular neutron port from the VM\u0027s perspective.\n\nI might be mistaken, but I believe this is how the current spec describes it.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"135ed97d6ed44e85cfdffd8e3826ba0bdb50898e","unresolved":false,"context_lines":[{"line_number":313,"context_line":"message (get_devices_details_list_and_failed_devices) to the Neutron server."},{"line_number":314,"context_line":"The details returned will have a new boolean field \"is_trunk\", to specify that"},{"line_number":315,"context_line":"the port is a trunk port and a field \"sub_ports\" to return all the details"},{"line_number":316,"context_line":"of the subports associated to this trunk. If the trunk port is not associated"},{"line_number":317,"context_line":"to any subport, the L2 agent will install a drop flow to drop all the traffic,"},{"line_number":318,"context_line":"since no traffic is supposed to flow when no subport is defined."},{"line_number":319,"context_line":"After that it will send a message notifying that the trunk port is up."},{"line_number":320,"context_line":"Neutron will send back to Nova an event to signal that the wiring was successful."},{"line_number":321,"context_line":""}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_f0154349","line":318,"range":{"start_line":316,"start_character":42,"end_line":318,"end_character":64},"in_reply_to":"bab6814e_9eececc8","updated":"2016-05-19 13:31:16.000000000","message":"The final version of the spec says a trunk create must refer to exactly one parent port. So a vm with a port cannot boot without the (vm-internal-)untagged network connectivity.\n\nThis applies to lines 362, 371 and 399 too.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"0834b467007206792fa912de8be12b376591c080","unresolved":false,"context_lines":[{"line_number":317,"context_line":"to any subport, the L2 agent will install a drop flow to drop all the traffic,"},{"line_number":318,"context_line":"since no traffic is supposed to flow when no subport is defined."},{"line_number":319,"context_line":"After that it will send a message notifying that the trunk port is up."},{"line_number":320,"context_line":"Neutron will send back to Nova an event to signal that the wiring was successful."},{"line_number":321,"context_line":""},{"line_number":322,"context_line":"Subport creation"},{"line_number":323,"context_line":"++++++++++++++++"}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_f86950fa","line":320,"updated":"2016-05-20 23:15:31.000000000","message":"Assume I do:\n\nneutron create-net foo\nneutron create-port foo\n# get hold of port id\nneutron trunk-create --parent port-id\nnova boot --nic port-id\u003dport-id\n\nThe question to answer is this: would a trunk without subports function correctly? I would think so. As a guest I should be able to have my interface connection on my foo network just like any other regular port.\n\nIs it useful? Probably not, as you\u0027d expect the the user to go and add subports later on.\n\nSo the other question is: do we want to allow for a successful boot off a trunk with at least one subport? If so, then we should probably fail the binding process rather than hitting the agent.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"9f825655d2b8121d74049ac5141c524026dd04bc","unresolved":false,"context_lines":[{"line_number":317,"context_line":"to any subport, the L2 agent will install a drop flow to drop all the traffic,"},{"line_number":318,"context_line":"since no traffic is supposed to flow when no subport is defined."},{"line_number":319,"context_line":"After that it will send a message notifying that the trunk port is up."},{"line_number":320,"context_line":"Neutron will send back to Nova an event to signal that the wiring was successful."},{"line_number":321,"context_line":""},{"line_number":322,"context_line":"Subport creation"},{"line_number":323,"context_line":"++++++++++++++++"}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_70b3774d","line":320,"in_reply_to":"bab6814e_ac8f5573","updated":"2016-05-23 17:30:43.000000000","message":"i the scenario above it would allow the tenant to add subports at a later date. so i think it is useful.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"08a44f4f5c0ef2600ce63b9b2d412255afecadeb","unresolved":false,"context_lines":[{"line_number":317,"context_line":"to any subport, the L2 agent will install a drop flow to drop all the traffic,"},{"line_number":318,"context_line":"since no traffic is supposed to flow when no subport is defined."},{"line_number":319,"context_line":"After that it will send a message notifying that the trunk port is up."},{"line_number":320,"context_line":"Neutron will send back to Nova an event to signal that the wiring was successful."},{"line_number":321,"context_line":""},{"line_number":322,"context_line":"Subport creation"},{"line_number":323,"context_line":"++++++++++++++++"}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_ac8f5573","line":320,"in_reply_to":"bab6814e_f86950fa","updated":"2016-05-23 15:08:05.000000000","message":"\u003e do we want to allow for a successful boot off a trunk with at least one subport?\n\nI think a boot using a trunk port with no subports should be successful. The user might not need to connect to multiple networks via subports at that point in time, but they might want to allow for the extra flexibility.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"b4da8f06477279f40c186847243451bd4f6c48da","unresolved":false,"context_lines":[{"line_number":322,"context_line":"Subport creation"},{"line_number":323,"context_line":"++++++++++++++++"},{"line_number":324,"context_line":""},{"line_number":325,"context_line":"When a subport is created or destroyed the L2 agent will get a notification."},{"line_number":326,"context_line":"Currently the L2 agent is notified only when a port is deleted, we need to add"},{"line_number":327,"context_line":"a notification if a subport is created. When the OVS agent receives a"},{"line_number":328,"context_line":"port_create notification it will create a vlan interface on the VM tap using"}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_945fab71","line":325,"updated":"2016-05-18 22:43:30.000000000","message":"I think we should be more clear here. Perhaps:\n\"When a subport is created or deleted on a trunk port...\"","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"808e23340e1d9899241b574767337836effcd200","unresolved":false,"context_lines":[{"line_number":322,"context_line":"Subport creation"},{"line_number":323,"context_line":"++++++++++++++++"},{"line_number":324,"context_line":""},{"line_number":325,"context_line":"When a subport is created or destroyed the L2 agent will get a notification."},{"line_number":326,"context_line":"Currently the L2 agent is notified only when a port is deleted, we need to add"},{"line_number":327,"context_line":"a notification if a subport is created. When the OVS agent receives a"},{"line_number":328,"context_line":"port_create notification it will create a vlan interface on the VM tap using"}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_9f0f9887","line":325,"in_reply_to":"bab6814e_38475878","updated":"2016-05-25 14:58:40.000000000","message":"OK","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"0834b467007206792fa912de8be12b376591c080","unresolved":false,"context_lines":[{"line_number":322,"context_line":"Subport creation"},{"line_number":323,"context_line":"++++++++++++++++"},{"line_number":324,"context_line":""},{"line_number":325,"context_line":"When a subport is created or destroyed the L2 agent will get a notification."},{"line_number":326,"context_line":"Currently the L2 agent is notified only when a port is deleted, we need to add"},{"line_number":327,"context_line":"a notification if a subport is created. When the OVS agent receives a"},{"line_number":328,"context_line":"port_create notification it will create a vlan interface on the VM tap using"}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_38475878","line":325,"in_reply_to":"bab6814e_945fab71","updated":"2016-05-20 23:15:31.000000000","message":"As I hinted before, the extent of how server/agent communication happens should be outside the scope of this patch. We\u0027ll take care of the RPC interface in some other patch and Ryan volunteered to help.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"135ed97d6ed44e85cfdffd8e3826ba0bdb50898e","unresolved":false,"context_lines":[{"line_number":324,"context_line":""},{"line_number":325,"context_line":"When a subport is created or destroyed the L2 agent will get a notification."},{"line_number":326,"context_line":"Currently the L2 agent is notified only when a port is deleted, we need to add"},{"line_number":327,"context_line":"a notification if a subport is created. When the OVS agent receives a"},{"line_number":328,"context_line":"port_create notification it will create a vlan interface on the VM tap using"},{"line_number":329,"context_line":"the VLAN id specified in the subport data."},{"line_number":330,"context_line":"Let\u0027s see how the models differ when using the iptables firewall or the ovs native"}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_27ee70ec","line":327,"range":{"start_line":327,"start_character":2,"end_line":327,"end_character":38},"updated":"2016-05-19 13:31:16.000000000","message":"I think it\u0027s would make sense to differentiate between subports created *before* the trunk port was plugged and its vif was created and *after*.\n\nThere\u0027s no point in sending notifications about subports with a trunk tap not yet existing.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"b4da8f06477279f40c186847243451bd4f6c48da","unresolved":false,"context_lines":[{"line_number":325,"context_line":"When a subport is created or destroyed the L2 agent will get a notification."},{"line_number":326,"context_line":"Currently the L2 agent is notified only when a port is deleted, we need to add"},{"line_number":327,"context_line":"a notification if a subport is created. When the OVS agent receives a"},{"line_number":328,"context_line":"port_create notification it will create a vlan interface on the VM tap using"},{"line_number":329,"context_line":"the VLAN id specified in the subport data."},{"line_number":330,"context_line":"Let\u0027s see how the models differ when using the iptables firewall or the ovs native"},{"line_number":331,"context_line":"firewall."}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_b49f4f02","line":328,"range":{"start_line":328,"start_character":0,"end_line":328,"end_character":11},"updated":"2016-05-18 22:43:30.000000000","message":"subport_create\n\nsubport_create should be distinct from a regular port_create, right?\n\nAlso, if a VM is created using a trunk port that contains multiple subports, does that mean multiple subport_create notifications will be sent to the agent? Or, should they be coalesced into a single notification, i.e. trunk_port_create, once a VM has been booted? Then adding individual subports to the trunk port after the VM has been booted would be subport_create notifications?","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"9f825655d2b8121d74049ac5141c524026dd04bc","unresolved":false,"context_lines":[{"line_number":325,"context_line":"When a subport is created or destroyed the L2 agent will get a notification."},{"line_number":326,"context_line":"Currently the L2 agent is notified only when a port is deleted, we need to add"},{"line_number":327,"context_line":"a notification if a subport is created. When the OVS agent receives a"},{"line_number":328,"context_line":"port_create notification it will create a vlan interface on the VM tap using"},{"line_number":329,"context_line":"the VLAN id specified in the subport data."},{"line_number":330,"context_line":"Let\u0027s see how the models differ when using the iptables firewall or the ovs native"},{"line_number":331,"context_line":"firewall."}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_90c7fb62","line":328,"range":{"start_line":328,"start_character":0,"end_line":328,"end_character":11},"in_reply_to":"bab6814e_7abda267","updated":"2016-05-23 17:30:43.000000000","message":"+1 this should probaly be split into a different section as it would be common to all options.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"808e23340e1d9899241b574767337836effcd200","unresolved":false,"context_lines":[{"line_number":325,"context_line":"When a subport is created or destroyed the L2 agent will get a notification."},{"line_number":326,"context_line":"Currently the L2 agent is notified only when a port is deleted, we need to add"},{"line_number":327,"context_line":"a notification if a subport is created. When the OVS agent receives a"},{"line_number":328,"context_line":"port_create notification it will create a vlan interface on the VM tap using"},{"line_number":329,"context_line":"the VLAN id specified in the subport data."},{"line_number":330,"context_line":"Let\u0027s see how the models differ when using the iptables firewall or the ovs native"},{"line_number":331,"context_line":"firewall."}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_9fa2f849","line":328,"range":{"start_line":328,"start_character":0,"end_line":328,"end_character":11},"in_reply_to":"bab6814e_90c7fb62","updated":"2016-05-25 14:58:40.000000000","message":"as Armando suggested we will have this discussion out of the scope of this spec","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"135ed97d6ed44e85cfdffd8e3826ba0bdb50898e","unresolved":false,"context_lines":[{"line_number":325,"context_line":"When a subport is created or destroyed the L2 agent will get a notification."},{"line_number":326,"context_line":"Currently the L2 agent is notified only when a port is deleted, we need to add"},{"line_number":327,"context_line":"a notification if a subport is created. When the OVS agent receives a"},{"line_number":328,"context_line":"port_create notification it will create a vlan interface on the VM tap using"},{"line_number":329,"context_line":"the VLAN id specified in the subport data."},{"line_number":330,"context_line":"Let\u0027s see how the models differ when using the iptables firewall or the ovs native"},{"line_number":331,"context_line":"firewall."}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_e7f588cb","line":328,"range":{"start_line":328,"start_character":0,"end_line":328,"end_character":11},"in_reply_to":"bab6814e_b49f4f02","updated":"2016-05-19 13:31:16.000000000","message":"Some subport creations can be coalesced, some cannot. For example it would be nice to wire all subports created before vm boot at once. But there\u0027s no chance of grouping subport creations happening in multiple api requests after vm boot.\n\nWithout coalescing at least some of the subport creations we could easily have performance problems.\n\nAn idea. Let\u0027s assume this workflow:\n\n neutron port-create net0 # port0\n neutron port-create net1 # port1\n neutron port-create net2 # port2\n neutron trunk-create \\\n     --port-id port0 \\\n     --subport port1,vlan,101 \\\n     --subport port2,vlan,102 # trunk0\n\n nova boot --nic port-id\u003dport0 ...\n\n neutron port-create net3 # port3\n neutron trunk-subport-add trunk0 \\\n     --subport port3,vlan,103\n\nThe actual wiring of port1 and port2 could be triggered by the plug() of port0. Communication between neutron-server and ovs-agent could happen in a single rpc request-response (eg. a subports field in the response to get_device_details()). There\u0027s no need to notify ovs-agent about the creation of port1 and port2.\n\nOn the other hand the wiring of port3 (and later) must be triggered by a new notification per port.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"366b54d9ab2de36836de9ac72dd794227f4bffca","unresolved":false,"context_lines":[{"line_number":325,"context_line":"When a subport is created or destroyed the L2 agent will get a notification."},{"line_number":326,"context_line":"Currently the L2 agent is notified only when a port is deleted, we need to add"},{"line_number":327,"context_line":"a notification if a subport is created. When the OVS agent receives a"},{"line_number":328,"context_line":"port_create notification it will create a vlan interface on the VM tap using"},{"line_number":329,"context_line":"the VLAN id specified in the subport data."},{"line_number":330,"context_line":"Let\u0027s see how the models differ when using the iptables firewall or the ovs native"},{"line_number":331,"context_line":"firewall."}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_7abda267","line":328,"range":{"start_line":328,"start_character":0,"end_line":328,"end_character":11},"in_reply_to":"bab6814e_e7f588cb","updated":"2016-05-19 16:32:58.000000000","message":"This is how I imagine the request flow as well.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"b4da8f06477279f40c186847243451bd4f6c48da","unresolved":false,"context_lines":[{"line_number":350,"context_line":"         |    port 1          port 2  |"},{"line_number":351,"context_line":"         |                            |"},{"line_number":352,"context_line":"         |           br-int           |"},{"line_number":353,"context_line":"         +----------------------------+"},{"line_number":354,"context_line":""},{"line_number":355,"context_line":"Let\u0027s assume the subport is on network1 and uses segmentation ID 100."},{"line_number":356,"context_line":"In the case of hybrid plugging the OVS agent will have to create the firewall"}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_0fb7283a","line":353,"updated":"2016-05-18 22:43:30.000000000","message":"These models are missing the defining feature of the \"vlan interface\" implementation: the actual vlan interfaces. In both models, there should be a vlan interface (which is created on the tap device corresponding to the VM\u0027s eth0) that matches the chosen vlan ID of the subport. That vlan interface would then be hooked into qbr2 (or br-int for the ovs-native firewall). Vlan 100-tagged traffic would be demuxed by the vlan interface (the tag is stripped automatically by the vlan interface), then forwarded (untagged) into qbr2 (or br-int for the ovs-native firewall), where the firewall rules can be applied to the untagged traffic.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":7448,"name":"Carl Baldwin","email":"carl@ecbaldwin.net","username":"carl-baldwin"},"change_message_id":"028ca8b8c48ce29031fe2b98c5445b25acfa7869","unresolved":false,"context_lines":[{"line_number":350,"context_line":"         |    port 1          port 2  |"},{"line_number":351,"context_line":"         |                            |"},{"line_number":352,"context_line":"         |           br-int           |"},{"line_number":353,"context_line":"         +----------------------------+"},{"line_number":354,"context_line":""},{"line_number":355,"context_line":"Let\u0027s assume the subport is on network1 and uses segmentation ID 100."},{"line_number":356,"context_line":"In the case of hybrid plugging the OVS agent will have to create the firewall"}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_c829978b","line":353,"in_reply_to":"bab6814e_0fb7283a","updated":"2016-05-18 23:37:09.000000000","message":"http://paste.openstack.org/show/497626/\n\nThis is more or less the way I see it.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"808e23340e1d9899241b574767337836effcd200","unresolved":false,"context_lines":[{"line_number":350,"context_line":"         |    port 1          port 2  |"},{"line_number":351,"context_line":"         |                            |"},{"line_number":352,"context_line":"         |           br-int           |"},{"line_number":353,"context_line":"         +----------------------------+"},{"line_number":354,"context_line":""},{"line_number":355,"context_line":"Let\u0027s assume the subport is on network1 and uses segmentation ID 100."},{"line_number":356,"context_line":"In the case of hybrid plugging the OVS agent will have to create the firewall"}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_1f8f28cd","line":353,"in_reply_to":"bab6814e_106cab06","updated":"2016-05-25 14:58:40.000000000","message":"thanks Carl!","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"366b54d9ab2de36836de9ac72dd794227f4bffca","unresolved":false,"context_lines":[{"line_number":350,"context_line":"         |    port 1          port 2  |"},{"line_number":351,"context_line":"         |                            |"},{"line_number":352,"context_line":"         |           br-int           |"},{"line_number":353,"context_line":"         +----------------------------+"},{"line_number":354,"context_line":""},{"line_number":355,"context_line":"Let\u0027s assume the subport is on network1 and uses segmentation ID 100."},{"line_number":356,"context_line":"In the case of hybrid plugging the OVS agent will have to create the firewall"}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_daac8e37","line":353,"in_reply_to":"bab6814e_c829978b","updated":"2016-05-19 16:32:58.000000000","message":"Yes, that is spot on.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"9f825655d2b8121d74049ac5141c524026dd04bc","unresolved":false,"context_lines":[{"line_number":350,"context_line":"         |    port 1          port 2  |"},{"line_number":351,"context_line":"         |                            |"},{"line_number":352,"context_line":"         |           br-int           |"},{"line_number":353,"context_line":"         +----------------------------+"},{"line_number":354,"context_line":""},{"line_number":355,"context_line":"Let\u0027s assume the subport is on network1 and uses segmentation ID 100."},{"line_number":356,"context_line":"In the case of hybrid plugging the OVS agent will have to create the firewall"}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_106cab06","line":353,"in_reply_to":"bab6814e_daac8e37","updated":"2016-05-23 17:30:43.000000000","message":"you probably also want to have a dashed line between\n\neth0 and eth0.100 but yes that looks good","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":7448,"name":"Carl Baldwin","email":"carl@ecbaldwin.net","username":"carl-baldwin"},"change_message_id":"028ca8b8c48ce29031fe2b98c5445b25acfa7869","unresolved":false,"context_lines":[{"line_number":357,"context_line":"bridge (qbr2), plug eth0.100 into it, connect it to br-int and set the subport"},{"line_number":358,"context_line":"ID in the external-ids of the port 2."},{"line_number":359,"context_line":""},{"line_number":360,"context_line":"*Inbound traffic*"},{"line_number":361,"context_line":""},{"line_number":362,"context_line":"The untagged traffic will flow from eth0 through qbr1. Since there\u0027s no subport"},{"line_number":363,"context_line":"with no segmentation ID set, the untagged traffic will be dropped."}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_0a7f16ab","line":360,"range":{"start_line":360,"start_character":1,"end_line":360,"end_character":16},"updated":"2016-05-18 23:37:09.000000000","message":"You mean inbound from the VM\u0027s point of view?  That\u0027s the assumption I make below.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"9f825655d2b8121d74049ac5141c524026dd04bc","unresolved":false,"context_lines":[{"line_number":357,"context_line":"bridge (qbr2), plug eth0.100 into it, connect it to br-int and set the subport"},{"line_number":358,"context_line":"ID in the external-ids of the port 2."},{"line_number":359,"context_line":""},{"line_number":360,"context_line":"*Inbound traffic*"},{"line_number":361,"context_line":""},{"line_number":362,"context_line":"The untagged traffic will flow from eth0 through qbr1. Since there\u0027s no subport"},{"line_number":363,"context_line":"with no segmentation ID set, the untagged traffic will be dropped."}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_5028f399","line":360,"range":{"start_line":360,"start_character":1,"end_line":360,"end_character":16},"in_reply_to":"bab6814e_0a7f16ab","updated":"2016-05-23 17:30:43.000000000","message":"+1 if we use inboud/outbound we need to be explcit about the context network or vm.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":7448,"name":"Carl Baldwin","email":"carl@ecbaldwin.net","username":"carl-baldwin"},"change_message_id":"028ca8b8c48ce29031fe2b98c5445b25acfa7869","unresolved":false,"context_lines":[{"line_number":362,"context_line":"The untagged traffic will flow from eth0 through qbr1. Since there\u0027s no subport"},{"line_number":363,"context_line":"with no segmentation ID set, the untagged traffic will be dropped."},{"line_number":364,"context_line":"For the traffic directed to port 2, the internal vlan ID of network1 will be"},{"line_number":365,"context_line":"stripped and vlan 100 will be pushed. The packet will then go through qbr2 where"},{"line_number":366,"context_line":"iptables rules will filter the traffic. *QUESTION* are firewall rules correctly"},{"line_number":367,"context_line":"applied by iptables when the traffic is tagged?"},{"line_number":368,"context_line":""}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_e8745b9d","line":365,"range":{"start_line":365,"start_character":13,"end_line":365,"end_character":36},"updated":"2016-05-18 23:37:09.000000000","message":"vlan 100 won\u0027t be pushed here.  The traffic will go through qbr2 untagged and then the vlan tag will get pushed by the tap1.100 interface.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"808e23340e1d9899241b574767337836effcd200","unresolved":false,"context_lines":[{"line_number":362,"context_line":"The untagged traffic will flow from eth0 through qbr1. Since there\u0027s no subport"},{"line_number":363,"context_line":"with no segmentation ID set, the untagged traffic will be dropped."},{"line_number":364,"context_line":"For the traffic directed to port 2, the internal vlan ID of network1 will be"},{"line_number":365,"context_line":"stripped and vlan 100 will be pushed. The packet will then go through qbr2 where"},{"line_number":366,"context_line":"iptables rules will filter the traffic. *QUESTION* are firewall rules correctly"},{"line_number":367,"context_line":"applied by iptables when the traffic is tagged?"},{"line_number":368,"context_line":""}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_5c4d3771","line":365,"range":{"start_line":365,"start_character":13,"end_line":365,"end_character":36},"in_reply_to":"bab6814e_e8745b9d","updated":"2016-05-25 14:58:40.000000000","message":"Done","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":7448,"name":"Carl Baldwin","email":"carl@ecbaldwin.net","username":"carl-baldwin"},"change_message_id":"028ca8b8c48ce29031fe2b98c5445b25acfa7869","unresolved":false,"context_lines":[{"line_number":363,"context_line":"with no segmentation ID set, the untagged traffic will be dropped."},{"line_number":364,"context_line":"For the traffic directed to port 2, the internal vlan ID of network1 will be"},{"line_number":365,"context_line":"stripped and vlan 100 will be pushed. The packet will then go through qbr2 where"},{"line_number":366,"context_line":"iptables rules will filter the traffic. *QUESTION* are firewall rules correctly"},{"line_number":367,"context_line":"applied by iptables when the traffic is tagged?"},{"line_number":368,"context_line":""},{"line_number":369,"context_line":"*Outbound traffic*"},{"line_number":370,"context_line":""}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_a86373db","line":367,"range":{"start_line":366,"start_character":40,"end_line":367,"end_character":47},"updated":"2016-05-18 23:37:09.000000000","message":"Question is moot with my response on L365.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"808e23340e1d9899241b574767337836effcd200","unresolved":false,"context_lines":[{"line_number":363,"context_line":"with no segmentation ID set, the untagged traffic will be dropped."},{"line_number":364,"context_line":"For the traffic directed to port 2, the internal vlan ID of network1 will be"},{"line_number":365,"context_line":"stripped and vlan 100 will be pushed. The packet will then go through qbr2 where"},{"line_number":366,"context_line":"iptables rules will filter the traffic. *QUESTION* are firewall rules correctly"},{"line_number":367,"context_line":"applied by iptables when the traffic is tagged?"},{"line_number":368,"context_line":""},{"line_number":369,"context_line":"*Outbound traffic*"},{"line_number":370,"context_line":""}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_3c50330a","line":367,"range":{"start_line":366,"start_character":40,"end_line":367,"end_character":47},"in_reply_to":"bab6814e_a86373db","updated":"2016-05-25 14:58:40.000000000","message":"Done","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"b4da8f06477279f40c186847243451bd4f6c48da","unresolved":false,"context_lines":[{"line_number":369,"context_line":"*Outbound traffic*"},{"line_number":370,"context_line":""},{"line_number":371,"context_line":"The untagged traffic will be dropped. Traffic tagged with vlan 100 will leave eth0.100,"},{"line_number":372,"context_line":"go through qbr2 where iptables rules will be applied and reach port 2. The OVS agent"},{"line_number":373,"context_line":"has installed a flow to strip vlan 100 and push the internal vlan of network1."},{"line_number":374,"context_line":""},{"line_number":375,"context_line":""},{"line_number":376,"context_line":"OVS Firewall case"}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_6f5eb4ea","line":373,"range":{"start_line":372,"start_character":71,"end_line":373,"end_character":77},"updated":"2016-05-18 22:43:30.000000000","message":"The use of vlan interfaces will not require this OVS flow. That\u0027s why this implementation is assumed to work for LinuxBridge.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"808e23340e1d9899241b574767337836effcd200","unresolved":false,"context_lines":[{"line_number":369,"context_line":"*Outbound traffic*"},{"line_number":370,"context_line":""},{"line_number":371,"context_line":"The untagged traffic will be dropped. Traffic tagged with vlan 100 will leave eth0.100,"},{"line_number":372,"context_line":"go through qbr2 where iptables rules will be applied and reach port 2. The OVS agent"},{"line_number":373,"context_line":"has installed a flow to strip vlan 100 and push the internal vlan of network1."},{"line_number":374,"context_line":""},{"line_number":375,"context_line":""},{"line_number":376,"context_line":"OVS Firewall case"}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_7a2be2eb","line":373,"range":{"start_line":372,"start_character":71,"end_line":373,"end_character":77},"in_reply_to":"bab6814e_08a9dfdf","updated":"2016-05-25 14:58:40.000000000","message":"Done","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":7448,"name":"Carl Baldwin","email":"carl@ecbaldwin.net","username":"carl-baldwin"},"change_message_id":"028ca8b8c48ce29031fe2b98c5445b25acfa7869","unresolved":false,"context_lines":[{"line_number":369,"context_line":"*Outbound traffic*"},{"line_number":370,"context_line":""},{"line_number":371,"context_line":"The untagged traffic will be dropped. Traffic tagged with vlan 100 will leave eth0.100,"},{"line_number":372,"context_line":"go through qbr2 where iptables rules will be applied and reach port 2. The OVS agent"},{"line_number":373,"context_line":"has installed a flow to strip vlan 100 and push the internal vlan of network1."},{"line_number":374,"context_line":""},{"line_number":375,"context_line":""},{"line_number":376,"context_line":"OVS Firewall case"}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_08a9dfdf","line":373,"range":{"start_line":372,"start_character":71,"end_line":373,"end_character":77},"in_reply_to":"bab6814e_6f5eb4ea","updated":"2016-05-18 23:37:09.000000000","message":"By the time the traffic gets here, the vlan 100 tagged has already been stripped.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"b4da8f06477279f40c186847243451bd4f6c48da","unresolved":false,"context_lines":[{"line_number":391,"context_line":"         |           br-int           |"},{"line_number":392,"context_line":"         +----------------------------+"},{"line_number":393,"context_line":""},{"line_number":394,"context_line":"When a subport is created the OVS agent will create the vlan interface eth0.100 and"},{"line_number":395,"context_line":"plug it into br-int."},{"line_number":396,"context_line":""},{"line_number":397,"context_line":"*Outbound traffic*"},{"line_number":398,"context_line":""}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_6f2cf44a","line":395,"range":{"start_line":394,"start_character":0,"end_line":395,"end_character":20},"updated":"2016-05-18 22:43:30.000000000","message":"The OVS agent will create the vlan interface on the VM\u0027s tap device that corresponds to eth0.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"d18809a5438e92881d1352f1211a539fed0f16c4","unresolved":false,"context_lines":[{"line_number":391,"context_line":"         |           br-int           |"},{"line_number":392,"context_line":"         +----------------------------+"},{"line_number":393,"context_line":""},{"line_number":394,"context_line":"When a subport is created the OVS agent will create the vlan interface eth0.100 and"},{"line_number":395,"context_line":"plug it into br-int."},{"line_number":396,"context_line":""},{"line_number":397,"context_line":"*Outbound traffic*"},{"line_number":398,"context_line":""}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_fc42a066","line":395,"range":{"start_line":394,"start_character":0,"end_line":395,"end_character":20},"in_reply_to":"bab6814e_638a9365","updated":"2016-05-23 18:42:40.000000000","message":"\u003e the tenant should be able to discover the trunk infomation via the metadata service from within the vm right.\n\nYes this is something that would be very useful to the VM, and it requires augmenting the Nova metadata service to fetch the metadata from the Neutron API.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"9f825655d2b8121d74049ac5141c524026dd04bc","unresolved":false,"context_lines":[{"line_number":391,"context_line":"         |           br-int           |"},{"line_number":392,"context_line":"         +----------------------------+"},{"line_number":393,"context_line":""},{"line_number":394,"context_line":"When a subport is created the OVS agent will create the vlan interface eth0.100 and"},{"line_number":395,"context_line":"plug it into br-int."},{"line_number":396,"context_line":""},{"line_number":397,"context_line":"*Outbound traffic*"},{"line_number":398,"context_line":""}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_638a9365","line":395,"range":{"start_line":394,"start_character":0,"end_line":395,"end_character":20},"in_reply_to":"bab6814e_6f2cf44a","updated":"2016-05-23 17:30:43.000000000","message":"yes the ovs agent cannot create interface inside the vm. that will have to be done by the tenant. the agent will create teh vlan subport of the tap as Rawlin pointed our. by the way i think we discussed this previosuly but the tenant should be able to discover the trunk infomation via the metadata service from within the vm right.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"9f825655d2b8121d74049ac5141c524026dd04bc","unresolved":false,"context_lines":[{"line_number":398,"context_line":""},{"line_number":399,"context_line":"The untagged traffic will flow and reach port 1 where it will be dropped. Traffic tagged"},{"line_number":400,"context_line":"with vlan 100 will leave eth0.100 reach port2. The firewall driver has installed flows to"},{"line_number":401,"context_line":"apply security group rules to port 2. The OVS agent will install flow rules to strip"},{"line_number":402,"context_line":"VLAN 100 and push the VLAN  used internally for network1."},{"line_number":403,"context_line":""},{"line_number":404,"context_line":"*Inbound traffic*"},{"line_number":405,"context_line":""}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_03b3c70a","line":402,"range":{"start_line":401,"start_character":38,"end_line":402,"end_character":57},"updated":"2016-05-23 17:30:43.000000000","message":"this is not needed. the tenants vlan will be striped by the tap subport and the br-int local vlan will be appled by the normal action when the untagged packet enters ovs via the tagged access port","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"808e23340e1d9899241b574767337836effcd200","unresolved":false,"context_lines":[{"line_number":398,"context_line":""},{"line_number":399,"context_line":"The untagged traffic will flow and reach port 1 where it will be dropped. Traffic tagged"},{"line_number":400,"context_line":"with vlan 100 will leave eth0.100 reach port2. The firewall driver has installed flows to"},{"line_number":401,"context_line":"apply security group rules to port 2. The OVS agent will install flow rules to strip"},{"line_number":402,"context_line":"VLAN 100 and push the VLAN  used internally for network1."},{"line_number":403,"context_line":""},{"line_number":404,"context_line":"*Inbound traffic*"},{"line_number":405,"context_line":""}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_1502670b","line":402,"range":{"start_line":401,"start_character":38,"end_line":402,"end_character":57},"in_reply_to":"bab6814e_03b3c70a","updated":"2016-05-25 14:58:40.000000000","message":"Done","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"9f825655d2b8121d74049ac5141c524026dd04bc","unresolved":false,"context_lines":[{"line_number":404,"context_line":"*Inbound traffic*"},{"line_number":405,"context_line":""},{"line_number":406,"context_line":"For port 1 all the traffic is dropped. The traffic for port 2 will be filtered by the rules"},{"line_number":407,"context_line":"installed by the firewall. The internal vlan ID of network1 will be stripped and vlan 100"},{"line_number":408,"context_line":"will be pushed."},{"line_number":409,"context_line":""},{"line_number":410,"context_line":"Subport deletion"}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_8375f732","line":407,"range":{"start_line":407,"start_character":68,"end_line":407,"end_character":76},"updated":"2016-05-23 17:30:43.000000000","message":"stripped by to normal action in ovs","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"808e23340e1d9899241b574767337836effcd200","unresolved":false,"context_lines":[{"line_number":404,"context_line":"*Inbound traffic*"},{"line_number":405,"context_line":""},{"line_number":406,"context_line":"For port 1 all the traffic is dropped. The traffic for port 2 will be filtered by the rules"},{"line_number":407,"context_line":"installed by the firewall. The internal vlan ID of network1 will be stripped and vlan 100"},{"line_number":408,"context_line":"will be pushed."},{"line_number":409,"context_line":""},{"line_number":410,"context_line":"Subport deletion"}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_90a2e549","line":407,"range":{"start_line":407,"start_character":68,"end_line":407,"end_character":76},"in_reply_to":"bab6814e_8375f732","updated":"2016-05-25 14:58:40.000000000","message":"Done","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"9f825655d2b8121d74049ac5141c524026dd04bc","unresolved":false,"context_lines":[{"line_number":405,"context_line":""},{"line_number":406,"context_line":"For port 1 all the traffic is dropped. The traffic for port 2 will be filtered by the rules"},{"line_number":407,"context_line":"installed by the firewall. The internal vlan ID of network1 will be stripped and vlan 100"},{"line_number":408,"context_line":"will be pushed."},{"line_number":409,"context_line":""},{"line_number":410,"context_line":"Subport deletion"},{"line_number":411,"context_line":"++++++++++++++++"}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_43a66fae","line":408,"range":{"start_line":408,"start_character":8,"end_line":408,"end_character":15},"updated":"2016-05-23 17:30:43.000000000","message":"pushed by the tap vlan subport.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"808e23340e1d9899241b574767337836effcd200","unresolved":false,"context_lines":[{"line_number":405,"context_line":""},{"line_number":406,"context_line":"For port 1 all the traffic is dropped. The traffic for port 2 will be filtered by the rules"},{"line_number":407,"context_line":"installed by the firewall. The internal vlan ID of network1 will be stripped and vlan 100"},{"line_number":408,"context_line":"will be pushed."},{"line_number":409,"context_line":""},{"line_number":410,"context_line":"Subport deletion"},{"line_number":411,"context_line":"++++++++++++++++"}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_d0846db2","line":408,"range":{"start_line":408,"start_character":8,"end_line":408,"end_character":15},"in_reply_to":"bab6814e_43a66fae","updated":"2016-05-25 14:58:40.000000000","message":"Done","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"0834b467007206792fa912de8be12b376591c080","unresolved":false,"context_lines":[{"line_number":410,"context_line":"Subport deletion"},{"line_number":411,"context_line":"++++++++++++++++"},{"line_number":412,"context_line":""},{"line_number":413,"context_line":"The OVS agent will get a notification that a subport needs to be deleted. The OVS agent"},{"line_number":414,"context_line":"will take care of removing the vlan interface, the firewall bridge if using iptables firewall"},{"line_number":415,"context_line":"and the port on br-int."},{"line_number":416,"context_line":""}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_18337415","line":413,"updated":"2016-05-20 23:15:31.000000000","message":"see line 320","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"808e23340e1d9899241b574767337836effcd200","unresolved":false,"context_lines":[{"line_number":410,"context_line":"Subport deletion"},{"line_number":411,"context_line":"++++++++++++++++"},{"line_number":412,"context_line":""},{"line_number":413,"context_line":"The OVS agent will get a notification that a subport needs to be deleted. The OVS agent"},{"line_number":414,"context_line":"will take care of removing the vlan interface, the firewall bridge if using iptables firewall"},{"line_number":415,"context_line":"and the port on br-int."},{"line_number":416,"context_line":""}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_5066bd29","line":413,"in_reply_to":"bab6814e_18337415","updated":"2016-05-25 14:58:40.000000000","message":"Done","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"135ed97d6ed44e85cfdffd8e3826ba0bdb50898e","unresolved":false,"context_lines":[{"line_number":418,"context_line":"++++++++++++"},{"line_number":419,"context_line":""},{"line_number":420,"context_line":"During resync the agent should check that all subports on br-int are still valid and"},{"line_number":421,"context_line":"delete the ones that are stale. The agent has no way of recognizing which port is a subport,"},{"line_number":422,"context_line":"it needs to get the device details from the server first and then perform the cleanup."},{"line_number":423,"context_line":""},{"line_number":424,"context_line":""},{"line_number":425,"context_line":"Multiple subports that carry untagged traffic"}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_4779f4f5","line":422,"range":{"start_line":421,"start_character":32,"end_line":422,"end_character":86},"updated":"2016-05-19 13:31:16.000000000","message":"What about storing the parent port\u0027s uuid as an external-id for subports? I think it would make the cascading cleanup after the deletion of a trunk port simpler. It would be nice not to ask neutron-server to see which subports are orphaned.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"9f825655d2b8121d74049ac5141c524026dd04bc","unresolved":false,"context_lines":[{"line_number":418,"context_line":"++++++++++++"},{"line_number":419,"context_line":""},{"line_number":420,"context_line":"During resync the agent should check that all subports on br-int are still valid and"},{"line_number":421,"context_line":"delete the ones that are stale. The agent has no way of recognizing which port is a subport,"},{"line_number":422,"context_line":"it needs to get the device details from the server first and then perform the cleanup."},{"line_number":423,"context_line":""},{"line_number":424,"context_line":""},{"line_number":425,"context_line":"Multiple subports that carry untagged traffic"}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_4342efae","line":422,"range":{"start_line":421,"start_character":32,"end_line":422,"end_character":86},"in_reply_to":"bab6814e_4779f4f5","updated":"2016-05-23 17:30:43.000000000","message":"+1 that should work","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"d18809a5438e92881d1352f1211a539fed0f16c4","unresolved":false,"context_lines":[{"line_number":418,"context_line":"++++++++++++"},{"line_number":419,"context_line":""},{"line_number":420,"context_line":"During resync the agent should check that all subports on br-int are still valid and"},{"line_number":421,"context_line":"delete the ones that are stale. The agent has no way of recognizing which port is a subport,"},{"line_number":422,"context_line":"it needs to get the device details from the server first and then perform the cleanup."},{"line_number":423,"context_line":""},{"line_number":424,"context_line":""},{"line_number":425,"context_line":"Multiple subports that carry untagged traffic"}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_b7eb1112","line":422,"range":{"start_line":421,"start_character":32,"end_line":422,"end_character":86},"in_reply_to":"bab6814e_4779f4f5","updated":"2016-05-23 18:42:40.000000000","message":"@Bence: Parent port of the subport, as in the port that was used to create the subport, not the trunk port\u0027s uuid, correct? I think that\u0027s what you\u0027re saying, because it makes sense to me.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"808e23340e1d9899241b574767337836effcd200","unresolved":false,"context_lines":[{"line_number":418,"context_line":"++++++++++++"},{"line_number":419,"context_line":""},{"line_number":420,"context_line":"During resync the agent should check that all subports on br-int are still valid and"},{"line_number":421,"context_line":"delete the ones that are stale. The agent has no way of recognizing which port is a subport,"},{"line_number":422,"context_line":"it needs to get the device details from the server first and then perform the cleanup."},{"line_number":423,"context_line":""},{"line_number":424,"context_line":""},{"line_number":425,"context_line":"Multiple subports that carry untagged traffic"}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_d6dd8671","line":422,"range":{"start_line":421,"start_character":32,"end_line":422,"end_character":86},"in_reply_to":"bab6814e_b7eb1112","updated":"2016-05-25 14:58:40.000000000","message":"Bence we can do that sure. I will put your suggestion in the trunk deletion paragraph. This is more about what the ovs agent should do in case of resync","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"6c9137baee05f6146dd33215b6c656aa3ccc75bf","unresolved":false,"context_lines":[{"line_number":418,"context_line":"++++++++++++"},{"line_number":419,"context_line":""},{"line_number":420,"context_line":"During resync the agent should check that all subports on br-int are still valid and"},{"line_number":421,"context_line":"delete the ones that are stale. The agent has no way of recognizing which port is a subport,"},{"line_number":422,"context_line":"it needs to get the device details from the server first and then perform the cleanup."},{"line_number":423,"context_line":""},{"line_number":424,"context_line":""},{"line_number":425,"context_line":"Multiple subports that carry untagged traffic"}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_3f7597ca","line":422,"range":{"start_line":421,"start_character":32,"end_line":422,"end_character":86},"in_reply_to":"bab6814e_d6dd8671","updated":"2016-05-26 12:00:49.000000000","message":"@Rossella: Sure. Better separate resnyc and deletion.\n\n@Rawlin: I mean each vif already contains the uuid of the port in the external_ids. In addition to that we could add another port uuid as a new external id to help decide if a vif belongs to a subport and if a subport is orphaned or not. Technically this id could be the uuid of the the trunk or the uuid of the parent port. If we prohibit the update of the port_id field of the trunk resource (do we?) then it\u0027s sufficient and simpler to store the parent port uuid.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"9f825655d2b8121d74049ac5141c524026dd04bc","unresolved":false,"context_lines":[{"line_number":425,"context_line":"Multiple subports that carry untagged traffic"},{"line_number":426,"context_line":"+++++++++++++++++++++++++++++++++++++++++++++"},{"line_number":427,"context_line":"WARNING!!"},{"line_number":428,"context_line":"With this model it\u0027s hard to support multiple subports that handle untagged traffic. Let\u0027s"},{"line_number":429,"context_line":"assume we have 2 subports, one connected to network1 and one to network2. They don\u0027t have"},{"line_number":430,"context_line":"the segmentation ID set, so they should carry untagged traffic. The agent should install"},{"line_number":431,"context_line":"flows for port 1 that tag the packets using the internal vlan id of network1. Since the"}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_c37b7ff3","line":428,"range":{"start_line":428,"start_character":0,"end_line":428,"end_character":85},"updated":"2016-05-23 17:30:43.000000000","message":"if i remember the spec correctly this is not allowed.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"b4da8f06477279f40c186847243451bd4f6c48da","unresolved":false,"context_lines":[{"line_number":426,"context_line":"+++++++++++++++++++++++++++++++++++++++++++++"},{"line_number":427,"context_line":"WARNING!!"},{"line_number":428,"context_line":"With this model it\u0027s hard to support multiple subports that handle untagged traffic. Let\u0027s"},{"line_number":429,"context_line":"assume we have 2 subports, one connected to network1 and one to network2. They don\u0027t have"},{"line_number":430,"context_line":"the segmentation ID set, so they should carry untagged traffic. The agent should install"},{"line_number":431,"context_line":"flows for port 1 that tag the packets using the internal vlan id of network1. Since the"},{"line_number":432,"context_line":"packet should also reach network2, it should be copied and then tagged using the internal"},{"line_number":433,"context_line":"vlan id assigned to network2."}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_2f778c3b","line":430,"range":{"start_line":429,"start_character":74,"end_line":430,"end_character":63},"updated":"2016-05-18 22:43:30.000000000","message":"I don\u0027t think subports without segmentation IDs will exist. By definition a subport has a segmentation_type and segmentation_id. Only the trunk port will carry untagged traffic.\n\nIf a VM needs to connect to multiple networks in an untagged manner, then the VM will need multiple trunk ports or multiple regular ports.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"0834b467007206792fa912de8be12b376591c080","unresolved":false,"context_lines":[{"line_number":427,"context_line":"WARNING!!"},{"line_number":428,"context_line":"With this model it\u0027s hard to support multiple subports that handle untagged traffic. Let\u0027s"},{"line_number":429,"context_line":"assume we have 2 subports, one connected to network1 and one to network2. They don\u0027t have"},{"line_number":430,"context_line":"the segmentation ID set, so they should carry untagged traffic. The agent should install"},{"line_number":431,"context_line":"flows for port 1 that tag the packets using the internal vlan id of network1. Since the"},{"line_number":432,"context_line":"packet should also reach network2, it should be copied and then tagged using the internal"},{"line_number":433,"context_line":"vlan id assigned to network2."}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_980e4439","line":430,"updated":"2016-05-20 23:15:31.000000000","message":"the model does not allow the specification of a subport without a segmentation id.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"135ed97d6ed44e85cfdffd8e3826ba0bdb50898e","unresolved":false,"context_lines":[{"line_number":426,"context_line":"+++++++++++++++++++++++++++++++++++++++++++++"},{"line_number":427,"context_line":"WARNING!!"},{"line_number":428,"context_line":"With this model it\u0027s hard to support multiple subports that handle untagged traffic. Let\u0027s"},{"line_number":429,"context_line":"assume we have 2 subports, one connected to network1 and one to network2. They don\u0027t have"},{"line_number":430,"context_line":"the segmentation ID set, so they should carry untagged traffic. The agent should install"},{"line_number":431,"context_line":"flows for port 1 that tag the packets using the internal vlan id of network1. Since the"},{"line_number":432,"context_line":"packet should also reach network2, it should be copied and then tagged using the internal"},{"line_number":433,"context_line":"vlan id assigned to network2."}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_466e1930","line":430,"range":{"start_line":429,"start_character":74,"end_line":430,"end_character":63},"in_reply_to":"bab6814e_2f778c3b","updated":"2016-05-19 13:31:16.000000000","message":"Yes, (vm-internal-)untagged traffic always belongs to the single trunk port. Subports must have (segmentation type, segmentation id), and this tuple must be unique in the scope of a trunk port.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"6c9137baee05f6146dd33215b6c656aa3ccc75bf","unresolved":false,"context_lines":[{"line_number":427,"context_line":"WARNING!!"},{"line_number":428,"context_line":"With this model it\u0027s hard to support multiple subports that handle untagged traffic. Let\u0027s"},{"line_number":429,"context_line":"assume we have 2 subports, one connected to network1 and one to network2. They don\u0027t have"},{"line_number":430,"context_line":"the segmentation ID set, so they should carry untagged traffic. The agent should install"},{"line_number":431,"context_line":"flows for port 1 that tag the packets using the internal vlan id of network1. Since the"},{"line_number":432,"context_line":"packet should also reach network2, it should be copied and then tagged using the internal"},{"line_number":433,"context_line":"vlan id assigned to network2."}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_bf1d0775","line":430,"in_reply_to":"bab6814e_63529336","updated":"2016-05-26 12:00:49.000000000","message":"In the final version there is no untagged subport. The untagged traffic belongs to the parent port. The parent port is a regular port. The parent port is always attached to a network.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"9f825655d2b8121d74049ac5141c524026dd04bc","unresolved":false,"context_lines":[{"line_number":427,"context_line":"WARNING!!"},{"line_number":428,"context_line":"With this model it\u0027s hard to support multiple subports that handle untagged traffic. Let\u0027s"},{"line_number":429,"context_line":"assume we have 2 subports, one connected to network1 and one to network2. They don\u0027t have"},{"line_number":430,"context_line":"the segmentation ID set, so they should carry untagged traffic. The agent should install"},{"line_number":431,"context_line":"flows for port 1 that tag the packets using the internal vlan id of network1. Since the"},{"line_number":432,"context_line":"packet should also reach network2, it should be copied and then tagged using the internal"},{"line_number":433,"context_line":"vlan id assigned to network2."}],"source_content_type":"text/x-rst","patch_set":1,"id":"bab6814e_63529336","line":430,"in_reply_to":"bab6814e_980e4439","updated":"2016-05-23 17:30:43.000000000","message":"how is an untaged subport currently modled.\n\nis it only the parent that can recieve untagged traffic or can you have a single subport that recives untagged traffic.\n\ni cant remember if the parent port is attached to a network or not. i believe it can be in which case the parent would be allowed to recived the untagged traffic.","commit_id":"84b95b113a9f5a2951b7f52ff979e308dfa7f943"},{"author":{"_account_id":7776,"name":"Paul Carver","email":"pcarver@att.com","username":"pcarver"},"change_message_id":"d5bb5f553faa37df42be5da9e7cff4ef31b86aab","unresolved":false,"context_lines":[{"line_number":306,"context_line":""},{"line_number":307,"context_line":"Trunk port creation"},{"line_number":308,"context_line":"+++++++++++++++++++"},{"line_number":309,"context_line":"A VM is spawn passing to Nova the port-id of a parent port associated with"},{"line_number":310,"context_line":"a trunk port. Nova will create the tap interface and will plug it into"},{"line_number":311,"context_line":"br-int or into the firewall bridge if using iptables firewall. In the"},{"line_number":312,"context_line":"external-ids of the port Nova will store the port ID of the trunk."}],"source_content_type":"text/x-rst","patch_set":2,"id":"bab6814e_05ccc4b9","line":309,"range":{"start_line":309,"start_character":8,"end_line":309,"end_character":13},"updated":"2016-05-26 12:26:03.000000000","message":"nit: spawn -\u003e spawned","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"40dc92ac0d9175b005fe32e57cbfb7f29954eaa0","unresolved":false,"context_lines":[{"line_number":306,"context_line":""},{"line_number":307,"context_line":"Trunk port creation"},{"line_number":308,"context_line":"+++++++++++++++++++"},{"line_number":309,"context_line":"A VM is spawn passing to Nova the port-id of a parent port associated with"},{"line_number":310,"context_line":"a trunk port. Nova will create the tap interface and will plug it into"},{"line_number":311,"context_line":"br-int or into the firewall bridge if using iptables firewall. In the"},{"line_number":312,"context_line":"external-ids of the port Nova will store the port ID of the trunk."}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_1845926f","line":309,"range":{"start_line":309,"start_character":8,"end_line":309,"end_character":13},"in_reply_to":"bab6814e_05ccc4b9","updated":"2016-05-30 16:16:14.000000000","message":"Done","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"767449f6fdf122986b24c9571285099dfed5235e","unresolved":false,"context_lines":[{"line_number":309,"context_line":"A VM is spawn passing to Nova the port-id of a parent port associated with"},{"line_number":310,"context_line":"a trunk port. Nova will create the tap interface and will plug it into"},{"line_number":311,"context_line":"br-int or into the firewall bridge if using iptables firewall. In the"},{"line_number":312,"context_line":"external-ids of the port Nova will store the port ID of the trunk."},{"line_number":313,"context_line":"The OVS agent detects that a new interface has been plugged. It gets"},{"line_number":314,"context_line":"the details of the new port and wires it."},{"line_number":315,"context_line":"The agent configures it in the same way as a traditional port: packets coming out"}],"source_content_type":"text/x-rst","patch_set":2,"id":"bab6814e_bc799595","line":312,"range":{"start_line":312,"start_character":60,"end_line":312,"end_character":65},"updated":"2016-05-26 15:35:48.000000000","message":"Saying \u0027... Nova will store the port ID of the parent port as usual.\u0027 would be clearer here. I guess that\u0027s what you meant. Nova should not and will not know about the trunk resource.","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"40dc92ac0d9175b005fe32e57cbfb7f29954eaa0","unresolved":false,"context_lines":[{"line_number":309,"context_line":"A VM is spawn passing to Nova the port-id of a parent port associated with"},{"line_number":310,"context_line":"a trunk port. Nova will create the tap interface and will plug it into"},{"line_number":311,"context_line":"br-int or into the firewall bridge if using iptables firewall. In the"},{"line_number":312,"context_line":"external-ids of the port Nova will store the port ID of the trunk."},{"line_number":313,"context_line":"The OVS agent detects that a new interface has been plugged. It gets"},{"line_number":314,"context_line":"the details of the new port and wires it."},{"line_number":315,"context_line":"The agent configures it in the same way as a traditional port: packets coming out"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_9807c21b","line":312,"range":{"start_line":312,"start_character":60,"end_line":312,"end_character":65},"in_reply_to":"bab6814e_bc799595","updated":"2016-05-30 16:16:14.000000000","message":"OK","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"767449f6fdf122986b24c9571285099dfed5235e","unresolved":false,"context_lines":[{"line_number":310,"context_line":"a trunk port. Nova will create the tap interface and will plug it into"},{"line_number":311,"context_line":"br-int or into the firewall bridge if using iptables firewall. In the"},{"line_number":312,"context_line":"external-ids of the port Nova will store the port ID of the trunk."},{"line_number":313,"context_line":"The OVS agent detects that a new interface has been plugged. It gets"},{"line_number":314,"context_line":"the details of the new port and wires it."},{"line_number":315,"context_line":"The agent configures it in the same way as a traditional port: packets coming out"},{"line_number":316,"context_line":"from the VM will be tagged using the internal vlan id associated to the network,"}],"source_content_type":"text/x-rst","patch_set":2,"id":"bab6814e_dcba99e0","line":313,"range":{"start_line":313,"start_character":33,"end_line":313,"end_character":42},"updated":"2016-05-26 15:35:48.000000000","message":"IMO \u0027vif\u0027 is the technically correct term.","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"40dc92ac0d9175b005fe32e57cbfb7f29954eaa0","unresolved":false,"context_lines":[{"line_number":310,"context_line":"a trunk port. Nova will create the tap interface and will plug it into"},{"line_number":311,"context_line":"br-int or into the firewall bridge if using iptables firewall. In the"},{"line_number":312,"context_line":"external-ids of the port Nova will store the port ID of the trunk."},{"line_number":313,"context_line":"The OVS agent detects that a new interface has been plugged. It gets"},{"line_number":314,"context_line":"the details of the new port and wires it."},{"line_number":315,"context_line":"The agent configures it in the same way as a traditional port: packets coming out"},{"line_number":316,"context_line":"from the VM will be tagged using the internal vlan id associated to the network,"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_fba888dd","line":313,"range":{"start_line":313,"start_character":33,"end_line":313,"end_character":42},"in_reply_to":"bab6814e_dcba99e0","updated":"2016-05-30 16:16:14.000000000","message":"I was omitting the adjective \"virtual\" since I assume the assumption is that we are dealing with virtual stuff...anyway let\u0027s change it if you feel it\u0027s not clear enough.","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":7776,"name":"Paul Carver","email":"pcarver@att.com","username":"pcarver"},"change_message_id":"d5bb5f553faa37df42be5da9e7cff4ef31b86aab","unresolved":false,"context_lines":[{"line_number":313,"context_line":"The OVS agent detects that a new interface has been plugged. It gets"},{"line_number":314,"context_line":"the details of the new port and wires it."},{"line_number":315,"context_line":"The agent configures it in the same way as a traditional port: packets coming out"},{"line_number":316,"context_line":"from the VM will be tagged using the internal vlan id associated to the network,"},{"line_number":317,"context_line":"packets going to the VM will be stripped of the vlan id."},{"line_number":318,"context_line":"After wiring it successfully the OVS agent will send a message notifying that the"},{"line_number":319,"context_line":"trunk port is up. Neutron will send back to Nova an event to signal that the"}],"source_content_type":"text/x-rst","patch_set":2,"id":"bab6814e_654030a6","line":316,"range":{"start_line":316,"start_character":46,"end_line":316,"end_character":50},"updated":"2016-05-26 12:26:03.000000000","message":"nit: VLAN really ought to be capitalized everywhere (except in code where Python namings standards dictate otherwise) because it\u0027s an acronym.","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"40dc92ac0d9175b005fe32e57cbfb7f29954eaa0","unresolved":false,"context_lines":[{"line_number":313,"context_line":"The OVS agent detects that a new interface has been plugged. It gets"},{"line_number":314,"context_line":"the details of the new port and wires it."},{"line_number":315,"context_line":"The agent configures it in the same way as a traditional port: packets coming out"},{"line_number":316,"context_line":"from the VM will be tagged using the internal vlan id associated to the network,"},{"line_number":317,"context_line":"packets going to the VM will be stripped of the vlan id."},{"line_number":318,"context_line":"After wiring it successfully the OVS agent will send a message notifying that the"},{"line_number":319,"context_line":"trunk port is up. Neutron will send back to Nova an event to signal that the"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_db82e451","line":316,"range":{"start_line":316,"start_character":46,"end_line":316,"end_character":50},"in_reply_to":"bab6814e_654030a6","updated":"2016-05-30 16:16:14.000000000","message":"you are right :) same for id","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"767449f6fdf122986b24c9571285099dfed5235e","unresolved":false,"context_lines":[{"line_number":315,"context_line":"The agent configures it in the same way as a traditional port: packets coming out"},{"line_number":316,"context_line":"from the VM will be tagged using the internal vlan id associated to the network,"},{"line_number":317,"context_line":"packets going to the VM will be stripped of the vlan id."},{"line_number":318,"context_line":"After wiring it successfully the OVS agent will send a message notifying that the"},{"line_number":319,"context_line":"trunk port is up. Neutron will send back to Nova an event to signal that the"},{"line_number":320,"context_line":"wiring was successful."},{"line_number":321,"context_line":""}],"source_content_type":"text/x-rst","patch_set":2,"id":"bab6814e_fcf4bdf1","line":318,"range":{"start_line":318,"start_character":55,"end_line":318,"end_character":72},"updated":"2016-05-26 15:35:48.000000000","message":"message notifying [neutron-server]","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"40dc92ac0d9175b005fe32e57cbfb7f29954eaa0","unresolved":false,"context_lines":[{"line_number":315,"context_line":"The agent configures it in the same way as a traditional port: packets coming out"},{"line_number":316,"context_line":"from the VM will be tagged using the internal vlan id associated to the network,"},{"line_number":317,"context_line":"packets going to the VM will be stripped of the vlan id."},{"line_number":318,"context_line":"After wiring it successfully the OVS agent will send a message notifying that the"},{"line_number":319,"context_line":"trunk port is up. Neutron will send back to Nova an event to signal that the"},{"line_number":320,"context_line":"wiring was successful."},{"line_number":321,"context_line":""}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_9be85c12","line":318,"range":{"start_line":318,"start_character":55,"end_line":318,"end_character":72},"in_reply_to":"bab6814e_fcf4bdf1","updated":"2016-05-30 16:16:14.000000000","message":"Done","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"c483fbdf68a2cd69e73b1e501d68e75c1e47212a","unresolved":false,"context_lines":[{"line_number":317,"context_line":"packets going to the VM will be stripped of the vlan id."},{"line_number":318,"context_line":"After wiring it successfully the OVS agent will send a message notifying that the"},{"line_number":319,"context_line":"trunk port is up. Neutron will send back to Nova an event to signal that the"},{"line_number":320,"context_line":"wiring was successful."},{"line_number":321,"context_line":""},{"line_number":322,"context_line":"Subport creation"},{"line_number":323,"context_line":"++++++++++++++++"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_d7c14b30","line":320,"updated":"2016-05-27 21:49:40.000000000","message":"If I may suggest, I think I see these scenarios:\n\n a) Boot a trunk with a parent and 0 subports.\n b) Boot a trunk with a parent and N subports.\n c) Add/remove a sub-port after boot.\n d) Add/remove subports before boot.\n e) Delete the trunk.\n\nI think you covered these overall, but for instance at line 322 I think you\u0027re implicitly assume scenario c) whereas scenario d) leads to nothing done agent side.\n\nPerhaps it\u0027s worth being more specific?","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"40dc92ac0d9175b005fe32e57cbfb7f29954eaa0","unresolved":false,"context_lines":[{"line_number":317,"context_line":"packets going to the VM will be stripped of the vlan id."},{"line_number":318,"context_line":"After wiring it successfully the OVS agent will send a message notifying that the"},{"line_number":319,"context_line":"trunk port is up. Neutron will send back to Nova an event to signal that the"},{"line_number":320,"context_line":"wiring was successful."},{"line_number":321,"context_line":""},{"line_number":322,"context_line":"Subport creation"},{"line_number":323,"context_line":"++++++++++++++++"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_db5e64e8","line":320,"in_reply_to":"9abb7d3a_d7c14b30","updated":"2016-05-30 16:16:14.000000000","message":"will do","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"e34bc7e7d197abfa5b064c1483f859f71d0566f8","unresolved":false,"context_lines":[{"line_number":354,"context_line":"         |           br-int           |"},{"line_number":355,"context_line":"         +----------------------------+"},{"line_number":356,"context_line":""},{"line_number":357,"context_line":"Let\u0027s assume the subport is on network1 and uses segmentation ID 100."},{"line_number":358,"context_line":"In the case of hybrid plugging the OVS agent will have to create the firewall"},{"line_number":359,"context_line":"bridge (qbr2), create tap1.100 and plug it into qbr2. It will connect qbr2 to"},{"line_number":360,"context_line":"br-int and set the subport ID in the external-ids of port 2."}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_6afeb66c","line":357,"range":{"start_line":357,"start_character":31,"end_line":357,"end_character":39},"updated":"2016-05-27 22:54:37.000000000","message":"I think it would make more sense if this was called \"network2\". Then it would match qbr2 and port 2.","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"e34bc7e7d197abfa5b064c1483f859f71d0566f8","unresolved":false,"context_lines":[{"line_number":373,"context_line":"firewall rules will be applied. Traffic tagged with vlan 100 will leave eth0.100,"},{"line_number":374,"context_line":"go through tap1.100 where the vlan 100 is stripped. It will reach qbr2 where"},{"line_number":375,"context_line":"iptables rules will be applied and go to port 2. The internal vlan of network1"},{"line_number":376,"context_line":"will be pushed by br-int when the packet enters port2 that it\u0027s a tagged port."},{"line_number":377,"context_line":""},{"line_number":378,"context_line":""},{"line_number":379,"context_line":"OVS Firewall case"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_2a794efd","line":376,"range":{"start_line":376,"start_character":54,"end_line":376,"end_character":58},"updated":"2016-05-27 22:54:37.000000000","message":"This sentence doesn\u0027t sound right. Did you mean \u0027that\u0027 -\u003e \u0027because\u0027?","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"40dc92ac0d9175b005fe32e57cbfb7f29954eaa0","unresolved":false,"context_lines":[{"line_number":373,"context_line":"firewall rules will be applied. Traffic tagged with vlan 100 will leave eth0.100,"},{"line_number":374,"context_line":"go through tap1.100 where the vlan 100 is stripped. It will reach qbr2 where"},{"line_number":375,"context_line":"iptables rules will be applied and go to port 2. The internal vlan of network1"},{"line_number":376,"context_line":"will be pushed by br-int when the packet enters port2 that it\u0027s a tagged port."},{"line_number":377,"context_line":""},{"line_number":378,"context_line":""},{"line_number":379,"context_line":"OVS Firewall case"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_5142553f","line":376,"range":{"start_line":376,"start_character":54,"end_line":376,"end_character":58},"in_reply_to":"9abb7d3a_2a794efd","updated":"2016-05-30 16:16:14.000000000","message":"yes :)","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"767449f6fdf122986b24c9571285099dfed5235e","unresolved":false,"context_lines":[{"line_number":386,"context_line":"         |   eth0            eth0.100 |"},{"line_number":387,"context_line":"         +-----+-----------------+----+"},{"line_number":388,"context_line":"               |                 |"},{"line_number":389,"context_line":"               |                 |"},{"line_number":390,"context_line":"           +---+---+       +-----+-----+"},{"line_number":391,"context_line":"           | tap1  |-------|  tap1.100 |"},{"line_number":392,"context_line":"           +---+---+       +-----+-----+"}],"source_content_type":"text/x-rst","patch_set":2,"id":"bab6814e_c2659245","line":389,"range":{"start_line":389,"start_character":33,"end_line":389,"end_character":34},"updated":"2016-05-26 15:35:48.000000000","message":"This line is here by accident, right?","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"40dc92ac0d9175b005fe32e57cbfb7f29954eaa0","unresolved":false,"context_lines":[{"line_number":386,"context_line":"         |   eth0            eth0.100 |"},{"line_number":387,"context_line":"         +-----+-----------------+----+"},{"line_number":388,"context_line":"               |                 |"},{"line_number":389,"context_line":"               |                 |"},{"line_number":390,"context_line":"           +---+---+       +-----+-----+"},{"line_number":391,"context_line":"           | tap1  |-------|  tap1.100 |"},{"line_number":392,"context_line":"           +---+---+       +-----+-----+"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_f1f6e9f7","line":389,"range":{"start_line":389,"start_character":33,"end_line":389,"end_character":34},"in_reply_to":"bab6814e_5b096f94","updated":"2016-05-30 16:16:14.000000000","message":"yes it\u0027s there by accident.\n@sean I think adding a line between eth0 and eth0.100 would make the draw more confusing","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"b0d9a56037faf99c1eced82bb7c255f513811f64","unresolved":false,"context_lines":[{"line_number":386,"context_line":"         |   eth0            eth0.100 |"},{"line_number":387,"context_line":"         +-----+-----------------+----+"},{"line_number":388,"context_line":"               |                 |"},{"line_number":389,"context_line":"               |                 |"},{"line_number":390,"context_line":"           +---+---+       +-----+-----+"},{"line_number":391,"context_line":"           | tap1  |-------|  tap1.100 |"},{"line_number":392,"context_line":"           +---+---+       +-----+-----+"}],"source_content_type":"text/x-rst","patch_set":2,"id":"bab6814e_5b096f94","line":389,"range":{"start_line":389,"start_character":33,"end_line":389,"end_character":34},"in_reply_to":"bab6814e_c2659245","updated":"2016-05-26 22:25:43.000000000","message":"there should be a line between eth0 and eth0.100\nbut not eth0.100 and tap1.100","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"c483fbdf68a2cd69e73b1e501d68e75c1e47212a","unresolved":false,"context_lines":[{"line_number":423,"context_line":""},{"line_number":424,"context_line":"When a trunk port is deleted, all the subports associated with that trunk should be"},{"line_number":425,"context_line":"deleted too. To recognize them, the parent port id might be stored in the subport"},{"line_number":426,"context_line":"external_ids."},{"line_number":427,"context_line":""},{"line_number":428,"context_line":"Subport deletion"},{"line_number":429,"context_line":"++++++++++++++++"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_17659344","line":426,"updated":"2016-05-27 21:49:40.000000000","message":"That\u0027s a good point. By the way, one of the spec constraints was that the deleting a port that is an active parent in a trunk should be forbidden. When Nova deletes a VM, it ends up deleting the associated VIFs if it has created the port, but since the port will be passed to Nova, Nova won\u0027t touch it (at least that\u0027s my recollection). So the cascade delete would only need to be executed when deleting the trunk, and I wonder if storing the trunk id is more appropriate (alongside port id if necessary).","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"40dc92ac0d9175b005fe32e57cbfb7f29954eaa0","unresolved":false,"context_lines":[{"line_number":423,"context_line":""},{"line_number":424,"context_line":"When a trunk port is deleted, all the subports associated with that trunk should be"},{"line_number":425,"context_line":"deleted too. To recognize them, the parent port id might be stored in the subport"},{"line_number":426,"context_line":"external_ids."},{"line_number":427,"context_line":""},{"line_number":428,"context_line":"Subport deletion"},{"line_number":429,"context_line":"++++++++++++++++"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_23e4a639","line":426,"in_reply_to":"9abb7d3a_17659344","updated":"2016-05-30 16:16:14.000000000","message":"I agree that is quicker to use the trunk id if the cascade delete will be triggered by the deletion of the trunk. I\u0027ve added more details here using your comment.","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"767449f6fdf122986b24c9571285099dfed5235e","unresolved":false,"context_lines":[{"line_number":452,"context_line":"               |"},{"line_number":453,"context_line":"         +-----+--------------------------+"},{"line_number":454,"context_line":"         |    tap1                        |"},{"line_number":455,"context_line":"         |            br-trunk            |"},{"line_number":456,"context_line":"         |                                |"},{"line_number":457,"context_line":"         | tp-patch-trunk  sp-patch-trunk |"},{"line_number":458,"context_line":"         +-----+-----------------+--------+"}],"source_content_type":"text/x-rst","patch_set":2,"id":"bab6814e_4d038a42","line":455,"range":{"start_line":455,"start_character":22,"end_line":455,"end_character":30},"updated":"2016-05-26 15:35:48.000000000","message":"There has to be one trunk bridge per trunk port. Please change the name to something that makes it obvious we\u0027ll have many, like: br-trunk-1 or tbr1...","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"b0d9a56037faf99c1eced82bb7c255f513811f64","unresolved":false,"context_lines":[{"line_number":452,"context_line":"               |"},{"line_number":453,"context_line":"         +-----+--------------------------+"},{"line_number":454,"context_line":"         |    tap1                        |"},{"line_number":455,"context_line":"         |            br-trunk            |"},{"line_number":456,"context_line":"         |                                |"},{"line_number":457,"context_line":"         | tp-patch-trunk  sp-patch-trunk |"},{"line_number":458,"context_line":"         +-----+-----------------+--------+"}],"source_content_type":"text/x-rst","patch_set":2,"id":"bab6814e_bb585b8b","line":455,"range":{"start_line":455,"start_character":22,"end_line":455,"end_character":30},"in_reply_to":"bab6814e_4d038a42","updated":"2016-05-26 22:25:43.000000000","message":"i would use tpb+13 characters of port uuid\ntrunk port bridge (tpb)","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"40dc92ac0d9175b005fe32e57cbfb7f29954eaa0","unresolved":false,"context_lines":[{"line_number":452,"context_line":"               |"},{"line_number":453,"context_line":"         +-----+--------------------------+"},{"line_number":454,"context_line":"         |    tap1                        |"},{"line_number":455,"context_line":"         |            br-trunk            |"},{"line_number":456,"context_line":"         |                                |"},{"line_number":457,"context_line":"         | tp-patch-trunk  sp-patch-trunk |"},{"line_number":458,"context_line":"         +-----+-----------------+--------+"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_a3453618","line":455,"range":{"start_line":455,"start_character":22,"end_line":455,"end_character":30},"in_reply_to":"bab6814e_bb585b8b","updated":"2016-05-30 16:16:14.000000000","message":"br-trunk-1 changed! 13 characters a bit too many for a diagram","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"e34bc7e7d197abfa5b064c1483f859f71d0566f8","unresolved":false,"context_lines":[{"line_number":454,"context_line":"         |    tap1                        |"},{"line_number":455,"context_line":"         |            br-trunk            |"},{"line_number":456,"context_line":"         |                                |"},{"line_number":457,"context_line":"         | tp-patch-trunk  sp-patch-trunk |"},{"line_number":458,"context_line":"         +-----+-----------------+--------+"},{"line_number":459,"context_line":"               |                 |"},{"line_number":460,"context_line":"               |                 |"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_ca2c02ae","line":457,"range":{"start_line":457,"start_character":27,"end_line":457,"end_character":41},"updated":"2016-05-27 22:54:37.000000000","message":"Similar to what @Bence and @sean mentioned on L455, I think these patch port names also need to use the port uuid postfix.","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"385d63353a1ecb1df7fc89179d3ae4a0d58f4b35","unresolved":false,"context_lines":[{"line_number":454,"context_line":"         |    tap1                        |"},{"line_number":455,"context_line":"         |            br-trunk            |"},{"line_number":456,"context_line":"         |                                |"},{"line_number":457,"context_line":"         | tp-patch-trunk  sp-patch-trunk |"},{"line_number":458,"context_line":"         +-----+-----------------+--------+"},{"line_number":459,"context_line":"               |                 |"},{"line_number":460,"context_line":"               |                 |"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_599d33bb","line":457,"range":{"start_line":457,"start_character":27,"end_line":457,"end_character":41},"in_reply_to":"9abb7d3a_83d47220","updated":"2016-05-31 10:53:38.000000000","message":"the dev ref is intended to show how its implemented though.\n\nits not evident that the uuid would be used in the implementation","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"40dc92ac0d9175b005fe32e57cbfb7f29954eaa0","unresolved":false,"context_lines":[{"line_number":454,"context_line":"         |    tap1                        |"},{"line_number":455,"context_line":"         |            br-trunk            |"},{"line_number":456,"context_line":"         |                                |"},{"line_number":457,"context_line":"         | tp-patch-trunk  sp-patch-trunk |"},{"line_number":458,"context_line":"         +-----+-----------------+--------+"},{"line_number":459,"context_line":"               |                 |"},{"line_number":460,"context_line":"               |                 |"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_83d47220","line":457,"range":{"start_line":457,"start_character":27,"end_line":457,"end_character":41},"in_reply_to":"9abb7d3a_ca2c02ae","updated":"2016-05-30 16:16:14.000000000","message":"I think using uuids is an overkill for this diagram that should help people get the big picture. It should be straightforward that it\u0027s one patch-trunk/patch-int pair per subport, isn\u0027t it?","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"cca7c99278c322e702f58efd58d0fae4ab1c6a72","unresolved":false,"context_lines":[{"line_number":454,"context_line":"         |    tap1                        |"},{"line_number":455,"context_line":"         |            br-trunk            |"},{"line_number":456,"context_line":"         |                                |"},{"line_number":457,"context_line":"         | tp-patch-trunk  sp-patch-trunk |"},{"line_number":458,"context_line":"         +-----+-----------------+--------+"},{"line_number":459,"context_line":"               |                 |"},{"line_number":460,"context_line":"               |                 |"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_e6210cb6","line":457,"range":{"start_line":457,"start_character":27,"end_line":457,"end_character":41},"in_reply_to":"9abb7d3a_ca2c02ae","updated":"2016-05-30 16:11:37.000000000","message":"hi yes.\novs requires that all port names must be unique for the ovs instance. they are not namespaced per bridge so they will have to use the uuid as part of there name\n\nmay i suggest the following nameing scheam\n\nbr-trunk \u003d\u003e \"tpb\"+\u003cuuid of parent\u003e[:13]\ntp-patch-trunk \u003d\u003e \"tpt\"+\u003cuuid of parent\u003e[:13]\nsp-patch-trunk \u003d\u003e \"spt\"+\u003cuuid of subport\u003e[:13]\ntp-patch-int \u003d\u003e \"tpi\"+\u003cuuid of parent\u003e[:13]\nsp-patch-int \u003d\u003e \"spi\"+\u003cuuid of subport\u003e[:13]","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"b0d9a56037faf99c1eced82bb7c255f513811f64","unresolved":false,"context_lines":[{"line_number":465,"context_line":"         |           br-int                |"},{"line_number":466,"context_line":"         +---------------------------------+"},{"line_number":467,"context_line":""},{"line_number":468,"context_line":"tp-patch-trunk: trunk bridge side of the patch port that implements the trunk port"},{"line_number":469,"context_line":"tp-patch-int: int bridge side of the patch port that implements the trunk port"},{"line_number":470,"context_line":"sp-patch-trunk: trunk bridge side of the patch port that implements the subport"},{"line_number":471,"context_line":"sp-patch-int: int bridge side of the patch port that implements the subport"},{"line_number":472,"context_line":""},{"line_number":473,"context_line":""},{"line_number":474,"context_line":"Trunk port creation"}],"source_content_type":"text/x-rst","patch_set":2,"id":"bab6814e_de654de1","line":471,"range":{"start_line":468,"start_character":0,"end_line":471,"end_character":75},"updated":"2016-05-26 22:25:43.000000000","message":"by the way incase anyone is still using use_veth_interconnection\u003dtrue these could be veth pairs instead.\n\nhttps://github.com/openstack/neutron/blob/master/neutron/plugins/ml2/drivers/openvswitch/agent/common/config.py#L65-L69\n\ncan we deprecate that option? if your using kernel ovs older then 2.0 you have bigger problems like the fact sfc,qos and conntrack/learn action security group drivers wont work because you dont have openflow 1.1 support let alone the 1.3+ they need.\n\nthis proposal will still work on pre 2.0 ovs though with veths its just as slow as the exiting veth pair in the hybrid plug configuration.","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"8d8d34ed51a9e606407c0d78a68916f5400deb11","unresolved":false,"context_lines":[{"line_number":465,"context_line":"         |           br-int                |"},{"line_number":466,"context_line":"         +---------------------------------+"},{"line_number":467,"context_line":""},{"line_number":468,"context_line":"tp-patch-trunk: trunk bridge side of the patch port that implements the trunk port"},{"line_number":469,"context_line":"tp-patch-int: int bridge side of the patch port that implements the trunk port"},{"line_number":470,"context_line":"sp-patch-trunk: trunk bridge side of the patch port that implements the subport"},{"line_number":471,"context_line":"sp-patch-int: int bridge side of the patch port that implements the subport"},{"line_number":472,"context_line":""},{"line_number":473,"context_line":""},{"line_number":474,"context_line":"Trunk port creation"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_d79bb970","line":471,"range":{"start_line":468,"start_character":0,"end_line":471,"end_character":75},"in_reply_to":"9abb7d3a_63cb4e1f","updated":"2016-05-31 08:02:11.000000000","message":"Done: https://bugs.launchpad.net/tempest/+bug/1587296","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"385d63353a1ecb1df7fc89179d3ae4a0d58f4b35","unresolved":false,"context_lines":[{"line_number":465,"context_line":"         |           br-int                |"},{"line_number":466,"context_line":"         +---------------------------------+"},{"line_number":467,"context_line":""},{"line_number":468,"context_line":"tp-patch-trunk: trunk bridge side of the patch port that implements the trunk port"},{"line_number":469,"context_line":"tp-patch-int: int bridge side of the patch port that implements the trunk port"},{"line_number":470,"context_line":"sp-patch-trunk: trunk bridge side of the patch port that implements the subport"},{"line_number":471,"context_line":"sp-patch-int: int bridge side of the patch port that implements the subport"},{"line_number":472,"context_line":""},{"line_number":473,"context_line":""},{"line_number":474,"context_line":"Trunk port creation"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_b9c9bf5b","line":471,"range":{"start_line":468,"start_character":0,"end_line":471,"end_character":75},"in_reply_to":"9abb7d3a_b7d275fe","updated":"2016-05-31 10:53:38.000000000","message":"actully it looks like there is some inconsistency between the release nodes and the code.\n\naccording to the release notes it is deprecated.\n\nhttps://github.com/openstack/neutron/blob/c7ca7c2a4e8e4ea67b170e99f914dab0fcf0dd29/releasenotes/source/locale/releasenotes.pot#L1157-L1163\n\nbut not according to the source\n\nhttps://github.com/openstack/neutron/blob/0eed94a777dbbdb2d8554cba136cd8e8145a9870/neutron/plugins/ml2/drivers/openvswitch/agent/common/config.py#L65-L69","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"3aa2bc36c7f8d0914393470b166a4cb638de0117","unresolved":false,"context_lines":[{"line_number":465,"context_line":"         |           br-int                |"},{"line_number":466,"context_line":"         +---------------------------------+"},{"line_number":467,"context_line":""},{"line_number":468,"context_line":"tp-patch-trunk: trunk bridge side of the patch port that implements the trunk port"},{"line_number":469,"context_line":"tp-patch-int: int bridge side of the patch port that implements the trunk port"},{"line_number":470,"context_line":"sp-patch-trunk: trunk bridge side of the patch port that implements the subport"},{"line_number":471,"context_line":"sp-patch-int: int bridge side of the patch port that implements the subport"},{"line_number":472,"context_line":""},{"line_number":473,"context_line":""},{"line_number":474,"context_line":"Trunk port creation"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_216bbef2","line":471,"range":{"start_line":468,"start_character":0,"end_line":471,"end_character":75},"in_reply_to":"9abb7d3a_b9c9bf5b","updated":"2016-05-31 17:58:47.000000000","message":"A bug had already been filed for this: https://bugs.launchpad.net/neutron/+bug/1550501\n\nThe option was deprecated to be removed in Ocata, but that patch was later reverted. See Neutron commit dab6070cafb95308cb8580856ecfdf213e8f77b8 . The review has some interesting comments about setting OVS internal ports to different namespaces (which is seen as harmful) vs using veth pairs between namespaces.\n\nThat said, I think changing the namespace of OVS internal ports might be the only area where we might actually need veth pairs instead. This comment - https://bugs.launchpad.net/neutron/+bug/1550501/comments/1 - makes it seem like we wouldn\u0027t need to require veth pair connections between trunk bridges and br-int.","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"88b31c9b4b59227861929770c1fd97de9f8d788a","unresolved":false,"context_lines":[{"line_number":465,"context_line":"         |           br-int                |"},{"line_number":466,"context_line":"         +---------------------------------+"},{"line_number":467,"context_line":""},{"line_number":468,"context_line":"tp-patch-trunk: trunk bridge side of the patch port that implements the trunk port"},{"line_number":469,"context_line":"tp-patch-int: int bridge side of the patch port that implements the trunk port"},{"line_number":470,"context_line":"sp-patch-trunk: trunk bridge side of the patch port that implements the subport"},{"line_number":471,"context_line":"sp-patch-int: int bridge side of the patch port that implements the subport"},{"line_number":472,"context_line":""},{"line_number":473,"context_line":""},{"line_number":474,"context_line":"Trunk port creation"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_b7d275fe","line":471,"range":{"start_line":468,"start_character":0,"end_line":471,"end_character":75},"in_reply_to":"9abb7d3a_d79bb970","updated":"2016-05-31 08:05:03.000000000","message":"I mean this: https://bugs.launchpad.net/neutron/+bug/1587296","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"40dc92ac0d9175b005fe32e57cbfb7f29954eaa0","unresolved":false,"context_lines":[{"line_number":465,"context_line":"         |           br-int                |"},{"line_number":466,"context_line":"         +---------------------------------+"},{"line_number":467,"context_line":""},{"line_number":468,"context_line":"tp-patch-trunk: trunk bridge side of the patch port that implements the trunk port"},{"line_number":469,"context_line":"tp-patch-int: int bridge side of the patch port that implements the trunk port"},{"line_number":470,"context_line":"sp-patch-trunk: trunk bridge side of the patch port that implements the subport"},{"line_number":471,"context_line":"sp-patch-int: int bridge side of the patch port that implements the subport"},{"line_number":472,"context_line":""},{"line_number":473,"context_line":""},{"line_number":474,"context_line":"Trunk port creation"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_63cb4e1f","line":471,"range":{"start_line":468,"start_character":0,"end_line":471,"end_character":75},"in_reply_to":"9abb7d3a_f8af457b","updated":"2016-05-30 16:16:14.000000000","message":"It\u0027s a valid point and I will add but I don\u0027t think the discussion about the deprecation of use_veth_interconnection belongs here. It\u0027s worth submitting a bug for it and discuss it there.","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"0680fc676222b9d702d1feb9bc7aa41e9a603d80","unresolved":false,"context_lines":[{"line_number":465,"context_line":"         |           br-int                |"},{"line_number":466,"context_line":"         +---------------------------------+"},{"line_number":467,"context_line":""},{"line_number":468,"context_line":"tp-patch-trunk: trunk bridge side of the patch port that implements the trunk port"},{"line_number":469,"context_line":"tp-patch-int: int bridge side of the patch port that implements the trunk port"},{"line_number":470,"context_line":"sp-patch-trunk: trunk bridge side of the patch port that implements the subport"},{"line_number":471,"context_line":"sp-patch-int: int bridge side of the patch port that implements the subport"},{"line_number":472,"context_line":""},{"line_number":473,"context_line":""},{"line_number":474,"context_line":"Trunk port creation"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_f8af457b","line":471,"range":{"start_line":468,"start_character":0,"end_line":471,"end_character":75},"in_reply_to":"bab6814e_de654de1","updated":"2016-05-30 13:21:47.000000000","message":"+1 for deprecating veth pairs between ovs bridges","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"767449f6fdf122986b24c9571285099dfed5235e","unresolved":false,"context_lines":[{"line_number":471,"context_line":"sp-patch-int: int bridge side of the patch port that implements the subport"},{"line_number":472,"context_line":""},{"line_number":473,"context_line":""},{"line_number":474,"context_line":"Trunk port creation"},{"line_number":475,"context_line":"+++++++++++++++++++"},{"line_number":476,"context_line":"A VM is spawn passing to Nova the port-id of a parent port associated with"},{"line_number":477,"context_line":"a trunk port. Nova will create the tap interface and will plug it into"}],"source_content_type":"text/x-rst","patch_set":2,"id":"bab6814e_0d32225e","line":474,"range":{"start_line":474,"start_character":0,"end_line":474,"end_character":19},"updated":"2016-05-26 15:35:48.000000000","message":"I think we are missing an aspect here. That is what vif types, drivers and vif details we\u0027ll have.\n\nI know this is an ovs-agent devref, but the agent and the vif driver together do one job. Design decisions made in one of them affect the other deeply.\n\nSo I suggest adding a section to each option about vif drivers. Answering these questions:\n\nIs some existing (especially ovs) vif driver sufficient without modification?\n\nDo we need to extend an existing vif driver?\n\nDo we need to add a new vif driver?\n\nWhere is boundary of responsibility between the vif driver and the neutron agent?\n\nAlso we should get feedback from nova (or os-vif-lib) folks.","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"40dc92ac0d9175b005fe32e57cbfb7f29954eaa0","unresolved":false,"context_lines":[{"line_number":471,"context_line":"sp-patch-int: int bridge side of the patch port that implements the subport"},{"line_number":472,"context_line":""},{"line_number":473,"context_line":""},{"line_number":474,"context_line":"Trunk port creation"},{"line_number":475,"context_line":"+++++++++++++++++++"},{"line_number":476,"context_line":"A VM is spawn passing to Nova the port-id of a parent port associated with"},{"line_number":477,"context_line":"a trunk port. Nova will create the tap interface and will plug it into"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_cc2187a1","line":474,"range":{"start_line":474,"start_character":0,"end_line":474,"end_character":19},"in_reply_to":"9abb7d3a_0d2bc483","updated":"2016-05-30 16:16:14.000000000","message":"I didn\u0027t add a section for that but I tried to capture what was discussed in the comments.","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"cca7c99278c322e702f58efd58d0fae4ab1c6a72","unresolved":false,"context_lines":[{"line_number":471,"context_line":"sp-patch-int: int bridge side of the patch port that implements the subport"},{"line_number":472,"context_line":""},{"line_number":473,"context_line":""},{"line_number":474,"context_line":"Trunk port creation"},{"line_number":475,"context_line":"+++++++++++++++++++"},{"line_number":476,"context_line":"A VM is spawn passing to Nova the port-id of a parent port associated with"},{"line_number":477,"context_line":"a trunk port. Nova will create the tap interface and will plug it into"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_09970149","line":474,"range":{"start_line":474,"start_character":0,"end_line":474,"end_character":19},"in_reply_to":"9abb7d3a_0d2bc483","updated":"2016-05-30 16:11:37.000000000","message":"i do not think we will need to create a new vif driver for this case. we should be able to extend the exising ovs plugin to cater for this.\n\nwe already handle both hybrid plug and non hybrid plug in the same plugin. im currently adding vhost-user support. bridge creation is a relatively minor change vs the logic currently used for  hybrid plug. after all its just\n\nsudo ovs-vsctl add-br tpbXYZ -- set bridge tpbXYZ external-ids:iface\u003dxyz\n\nall the plug driver will have to do is check if bridge exists. create it if it dose not. add port to bridge.\n\nthis can all resided within the ovs plugin https://github.com/openstack/os-vif/blob/master/vif_plug_ovs","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"e34bc7e7d197abfa5b064c1483f859f71d0566f8","unresolved":false,"context_lines":[{"line_number":471,"context_line":"sp-patch-int: int bridge side of the patch port that implements the subport"},{"line_number":472,"context_line":""},{"line_number":473,"context_line":""},{"line_number":474,"context_line":"Trunk port creation"},{"line_number":475,"context_line":"+++++++++++++++++++"},{"line_number":476,"context_line":"A VM is spawn passing to Nova the port-id of a parent port associated with"},{"line_number":477,"context_line":"a trunk port. Nova will create the tap interface and will plug it into"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_0d2bc483","line":474,"range":{"start_line":474,"start_character":0,"end_line":474,"end_character":19},"in_reply_to":"bab6814e_0d32225e","updated":"2016-05-27 22:54:37.000000000","message":"I agree that we need some discussion around this. I think option A can use the existing vif drivers without any modification and doesn\u0027t require Nova to have any special responsibilities.\n\nOption C will require new VIF drivers, meaning Nova/os-vif should have the responsibility of creating the OVS trunk bridge and the 1st patch port pair to br-int. Like this: https://review.openstack.org/#/c/321253/","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":7776,"name":"Paul Carver","email":"pcarver@att.com","username":"pcarver"},"change_message_id":"d5bb5f553faa37df42be5da9e7cff4ef31b86aab","unresolved":false,"context_lines":[{"line_number":473,"context_line":""},{"line_number":474,"context_line":"Trunk port creation"},{"line_number":475,"context_line":"+++++++++++++++++++"},{"line_number":476,"context_line":"A VM is spawn passing to Nova the port-id of a parent port associated with"},{"line_number":477,"context_line":"a trunk port. Nova will create the tap interface and will plug it into"},{"line_number":478,"context_line":"br-trunk. The OVS agent will be monitoring all the trunk bridges on the host"},{"line_number":479,"context_line":"(currently it monitors only br-int and the ancillary bridges). When it detects"}],"source_content_type":"text/x-rst","patch_set":2,"id":"bab6814e_858d54d5","line":476,"range":{"start_line":476,"start_character":8,"end_line":476,"end_character":13},"updated":"2016-05-26 12:26:03.000000000","message":"nit: spawn -\u003e spawned","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"40dc92ac0d9175b005fe32e57cbfb7f29954eaa0","unresolved":false,"context_lines":[{"line_number":473,"context_line":""},{"line_number":474,"context_line":"Trunk port creation"},{"line_number":475,"context_line":"+++++++++++++++++++"},{"line_number":476,"context_line":"A VM is spawn passing to Nova the port-id of a parent port associated with"},{"line_number":477,"context_line":"a trunk port. Nova will create the tap interface and will plug it into"},{"line_number":478,"context_line":"br-trunk. The OVS agent will be monitoring all the trunk bridges on the host"},{"line_number":479,"context_line":"(currently it monitors only br-int and the ancillary bridges). When it detects"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_a63f04d5","line":476,"range":{"start_line":476,"start_character":8,"end_line":476,"end_character":13},"in_reply_to":"bab6814e_858d54d5","updated":"2016-05-30 16:16:14.000000000","message":"Done","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"767449f6fdf122986b24c9571285099dfed5235e","unresolved":false,"context_lines":[{"line_number":474,"context_line":"Trunk port creation"},{"line_number":475,"context_line":"+++++++++++++++++++"},{"line_number":476,"context_line":"A VM is spawn passing to Nova the port-id of a parent port associated with"},{"line_number":477,"context_line":"a trunk port. Nova will create the tap interface and will plug it into"},{"line_number":478,"context_line":"br-trunk. The OVS agent will be monitoring all the trunk bridges on the host"},{"line_number":479,"context_line":"(currently it monitors only br-int and the ancillary bridges). When it detects"},{"line_number":480,"context_line":"that a new interface has been plugged into a trunk bridge, it will do the"}],"source_content_type":"text/x-rst","patch_set":2,"id":"bab6814e_c87928fe","line":477,"range":{"start_line":477,"start_character":14,"end_line":477,"end_character":30},"updated":"2016-05-26 15:35:48.000000000","message":"Another thing we have to talk about here is whether trunk ports are treated special or not. If we provide a new kind of wiring for trunk ports we have to choose between these alternatives:\n\n1) Trunk ports are wired differently than regular ports.\n\nPros:\n\n* No upgrade between different wirings.\n\nCons:\n\n* A regular port cannot be turned into a trunk port after a vm was booted on it. This reaches back to when certain API calls are allowed. In the spec we left this open. But a user could choose to boot a vm an a trunk port without subports to keep the chance of adding subports later.\n\n* Depending on where we draw the boundary between the agent and the vif driver, the vif driver may need to know about if a port is regular port or a trunk port. For example by using a different vif type or vif details.\n\n* Two code paths to maintain in the vif drivers and in the agent.\n\n2) All ports are wired the same way that is sufficient for trunk ports too.\n\nPros and cons: Just the other way around as in (1). We have to figure how to upgrade.","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"0680fc676222b9d702d1feb9bc7aa41e9a603d80","unresolved":false,"context_lines":[{"line_number":474,"context_line":"Trunk port creation"},{"line_number":475,"context_line":"+++++++++++++++++++"},{"line_number":476,"context_line":"A VM is spawn passing to Nova the port-id of a parent port associated with"},{"line_number":477,"context_line":"a trunk port. Nova will create the tap interface and will plug it into"},{"line_number":478,"context_line":"br-trunk. The OVS agent will be monitoring all the trunk bridges on the host"},{"line_number":479,"context_line":"(currently it monitors only br-int and the ancillary bridges). When it detects"},{"line_number":480,"context_line":"that a new interface has been plugged into a trunk bridge, it will do the"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_4ec0f92a","line":477,"in_reply_to":"9abb7d3a_57d6fbc5","updated":"2016-05-30 13:21:47.000000000","message":"I\u0027m not sure it\u0027s clear from the base patch we made a decision. That\u0027s what I found: \"if two separate code paths are assumed to be maintained in the OVS agent ...\" (line 231) Maybe it is worth rephrasing this to be clear-cut?","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"c483fbdf68a2cd69e73b1e501d68e75c1e47212a","unresolved":false,"context_lines":[{"line_number":474,"context_line":"Trunk port creation"},{"line_number":475,"context_line":"+++++++++++++++++++"},{"line_number":476,"context_line":"A VM is spawn passing to Nova the port-id of a parent port associated with"},{"line_number":477,"context_line":"a trunk port. Nova will create the tap interface and will plug it into"},{"line_number":478,"context_line":"br-trunk. The OVS agent will be monitoring all the trunk bridges on the host"},{"line_number":479,"context_line":"(currently it monitors only br-int and the ancillary bridges). When it detects"},{"line_number":480,"context_line":"that a new interface has been plugged into a trunk bridge, it will do the"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_57d6fbc5","line":477,"in_reply_to":"bab6814e_c87928fe","updated":"2016-05-27 21:49:40.000000000","message":"I think we already discussed this, and we are going with 1. See base patch.","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":21273,"name":"Ilya Chukhnakov","email":"Ilya.chukhnakov@gmail.com","username":"ichukhnakov"},"change_message_id":"0525c8caf23c5ed4029cc40ccb86e7a952f1709f","unresolved":false,"context_lines":[{"line_number":475,"context_line":"+++++++++++++++++++"},{"line_number":476,"context_line":"A VM is spawn passing to Nova the port-id of a parent port associated with"},{"line_number":477,"context_line":"a trunk port. Nova will create the tap interface and will plug it into"},{"line_number":478,"context_line":"br-trunk. The OVS agent will be monitoring all the trunk bridges on the host"},{"line_number":479,"context_line":"(currently it monitors only br-int and the ancillary bridges). When it detects"},{"line_number":480,"context_line":"that a new interface has been plugged into a trunk bridge, it will do the"},{"line_number":481,"context_line":"following:"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_f0d3995d","line":478,"range":{"start_line":478,"start_character":14,"end_line":478,"end_character":64},"updated":"2016-05-30 06:35:11.000000000","message":"I think it should be clarified whether the presence of multiple trunk bridges would or would not affect agent performance and scaling in case we have many trunk ports on the same host.","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"c483fbdf68a2cd69e73b1e501d68e75c1e47212a","unresolved":false,"context_lines":[{"line_number":475,"context_line":"+++++++++++++++++++"},{"line_number":476,"context_line":"A VM is spawn passing to Nova the port-id of a parent port associated with"},{"line_number":477,"context_line":"a trunk port. Nova will create the tap interface and will plug it into"},{"line_number":478,"context_line":"br-trunk. The OVS agent will be monitoring all the trunk bridges on the host"},{"line_number":479,"context_line":"(currently it monitors only br-int and the ancillary bridges). When it detects"},{"line_number":480,"context_line":"that a new interface has been plugged into a trunk bridge, it will do the"},{"line_number":481,"context_line":"following:"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_77b69f8b","line":478,"updated":"2016-05-27 21:49:40.000000000","message":"we\u0027d need to clarify here who expect to create the bridge, but so far so good.","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"40dc92ac0d9175b005fe32e57cbfb7f29954eaa0","unresolved":false,"context_lines":[{"line_number":475,"context_line":"+++++++++++++++++++"},{"line_number":476,"context_line":"A VM is spawn passing to Nova the port-id of a parent port associated with"},{"line_number":477,"context_line":"a trunk port. Nova will create the tap interface and will plug it into"},{"line_number":478,"context_line":"br-trunk. The OVS agent will be monitoring all the trunk bridges on the host"},{"line_number":479,"context_line":"(currently it monitors only br-int and the ancillary bridges). When it detects"},{"line_number":480,"context_line":"that a new interface has been plugged into a trunk bridge, it will do the"},{"line_number":481,"context_line":"following:"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_09df016a","line":478,"range":{"start_line":478,"start_character":14,"end_line":478,"end_character":64},"in_reply_to":"9abb7d3a_f0d3995d","updated":"2016-05-30 16:16:14.000000000","message":"@Armando will clarify\n@Ilya I don\u0027t expect the performance to degrade","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"767449f6fdf122986b24c9571285099dfed5235e","unresolved":false,"context_lines":[{"line_number":475,"context_line":"+++++++++++++++++++"},{"line_number":476,"context_line":"A VM is spawn passing to Nova the port-id of a parent port associated with"},{"line_number":477,"context_line":"a trunk port. Nova will create the tap interface and will plug it into"},{"line_number":478,"context_line":"br-trunk. The OVS agent will be monitoring all the trunk bridges on the host"},{"line_number":479,"context_line":"(currently it monitors only br-int and the ancillary bridges). When it detects"},{"line_number":480,"context_line":"that a new interface has been plugged into a trunk bridge, it will do the"},{"line_number":481,"context_line":"following:"},{"line_number":482,"context_line":" "},{"line_number":483,"context_line":"ovs-vsctl add-port br-trunk tp-patch-trunk -- set Interface tp-patch-trunk type\u003dpatch options:peer\u003dtp-patch-int"},{"line_number":484,"context_line":"ovs-vsctl add-port br-int tp-patch-int tag\u003d3 -- set Interface tp-patch-int type\u003dpatch options:peer\u003dtp-patch-trunk"}],"source_content_type":"text/x-rst","patch_set":2,"id":"bab6814e_ee69958d","line":481,"range":{"start_line":478,"start_character":10,"end_line":481,"end_character":10},"updated":"2016-05-26 15:35:48.000000000","message":"I have an alternative:\n\n* The vif driver creates the trunk bridge in plug().\n* The vif driver deletes the trunk bridge in unplug().\n* The vif driver plugs the patch cable for the trunk port.\n* The agent plugs the patch cables for all subports.\n* The agent monitors br-int only. It does not monitor the trunk bridges.\n* External_ids are added to ports on br-int, not to ports on the trunk bridges.\n\nMy reasons to suggest this:\n\n* This is analog to how the vif driver creates the linux bridges for iptables today.\n* More importantly: Nova does not wait for anybody before it plugs. Therefore the agent does not have a chance to create the trunk bridge before nova comes and plugs.","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"3aa2bc36c7f8d0914393470b166a4cb638de0117","unresolved":false,"context_lines":[{"line_number":478,"context_line":"br-trunk. The OVS agent will be monitoring all the trunk bridges on the host"},{"line_number":479,"context_line":"(currently it monitors only br-int and the ancillary bridges). When it detects"},{"line_number":480,"context_line":"that a new interface has been plugged into a trunk bridge, it will do the"},{"line_number":481,"context_line":"following:"},{"line_number":482,"context_line":" "},{"line_number":483,"context_line":"ovs-vsctl add-port br-trunk tp-patch-trunk -- set Interface tp-patch-trunk type\u003dpatch options:peer\u003dtp-patch-int"},{"line_number":484,"context_line":"ovs-vsctl add-port br-int tp-patch-int tag\u003d3 -- set Interface tp-patch-int type\u003dpatch options:peer\u003dtp-patch-trunk"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_41da42e9","line":481,"in_reply_to":"9abb7d3a_2c0013aa","updated":"2016-05-31 17:58:47.000000000","message":"I guess my concern wasn\u0027t that it would be difficult to monitor a large number of bridges on the host, but rather that having the OVS agent create the 1st patch port pair goes somewhat against its current workflow of polling for new ports on br-int.\n\n\u003e so i agree that os-vif/nova should create the bridge. the reason i dont want to have nova/os-vif create the patch port is so we can deprecate and then remove https://github.com/openstack/nova/blob/master/nova/conf/neutron.py#L32-L35.\n\nI think we will still be able to deprecate and remove that option with os-vif/nova creating the 1st patch port pair. Neutron can currently specify which bridge to plug into via the vif details, and Nova will honor that. In this case \u0027br-int\u0027 would be passed so the vif driver knows to plug the trunk bridge it creates into br-int. For kernel OVS, the vif driver will overwrite this value in the VIF object with the name of the trunk bridge it created for the libvirt config xml file (so that the tap is placed on the trunk bridge). This is completely analogous to the hybrid_ovs strategy today.\n\nIn terms of the \"use_veth_interconnection\" option, this comment from Assaf Muller - https://bugs.launchpad.net/neutron/+bug/1550501/comments/1 - makes it seem like using veth pairs instead of patch ports here would no longer be necessary for compatibility with older distros.","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"0680fc676222b9d702d1feb9bc7aa41e9a603d80","unresolved":false,"context_lines":[{"line_number":478,"context_line":"br-trunk. The OVS agent will be monitoring all the trunk bridges on the host"},{"line_number":479,"context_line":"(currently it monitors only br-int and the ancillary bridges). When it detects"},{"line_number":480,"context_line":"that a new interface has been plugged into a trunk bridge, it will do the"},{"line_number":481,"context_line":"following:"},{"line_number":482,"context_line":" "},{"line_number":483,"context_line":"ovs-vsctl add-port br-trunk tp-patch-trunk -- set Interface tp-patch-trunk type\u003dpatch options:peer\u003dtp-patch-int"},{"line_number":484,"context_line":"ovs-vsctl add-port br-int tp-patch-int tag\u003d3 -- set Interface tp-patch-int type\u003dpatch options:peer\u003dtp-patch-trunk"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_b8a446b9","line":481,"in_reply_to":"9abb7d3a_4df82ce6","updated":"2016-05-30 13:21:47.000000000","message":"I parsed sean\u0027s proposal like this:\n\nhttps://etherpad.openstack.org/p/vif-agent-division\n\n@Armando: \"bridge deletion should be probably taken care of during trunk deletion\" plus after trunk port re/un-bind on the old host.\n\n@Rawlin: Since \u0027ovsdb-client monitor\u0027 can monitor the total content of the ovs db, I don\u0027t think it really matters how many ovs objects we monitor.\n\nIf the vif driver creates the trunk bridge and neutron agent deletes it, I guess that also means we keep the trunk bridge around for example during nova stop/start.","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"cca7c99278c322e702f58efd58d0fae4ab1c6a72","unresolved":false,"context_lines":[{"line_number":478,"context_line":"br-trunk. The OVS agent will be monitoring all the trunk bridges on the host"},{"line_number":479,"context_line":"(currently it monitors only br-int and the ancillary bridges). When it detects"},{"line_number":480,"context_line":"that a new interface has been plugged into a trunk bridge, it will do the"},{"line_number":481,"context_line":"following:"},{"line_number":482,"context_line":" "},{"line_number":483,"context_line":"ovs-vsctl add-port br-trunk tp-patch-trunk -- set Interface tp-patch-trunk type\u003dpatch options:peer\u003dtp-patch-int"},{"line_number":484,"context_line":"ovs-vsctl add-port br-int tp-patch-int tag\u003d3 -- set Interface tp-patch-int type\u003dpatch options:peer\u003dtp-patch-trunk"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_e692ac9e","line":481,"in_reply_to":"9abb7d3a_b8a446b9","updated":"2016-05-30 16:11:37.000000000","message":"@bence yes that correlates well to what i was thinking.\nthanks for translating my rambling :)","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"40dc92ac0d9175b005fe32e57cbfb7f29954eaa0","unresolved":false,"context_lines":[{"line_number":478,"context_line":"br-trunk. The OVS agent will be monitoring all the trunk bridges on the host"},{"line_number":479,"context_line":"(currently it monitors only br-int and the ancillary bridges). When it detects"},{"line_number":480,"context_line":"that a new interface has been plugged into a trunk bridge, it will do the"},{"line_number":481,"context_line":"following:"},{"line_number":482,"context_line":" "},{"line_number":483,"context_line":"ovs-vsctl add-port br-trunk tp-patch-trunk -- set Interface tp-patch-trunk type\u003dpatch options:peer\u003dtp-patch-int"},{"line_number":484,"context_line":"ovs-vsctl add-port br-int tp-patch-int tag\u003d3 -- set Interface tp-patch-int type\u003dpatch options:peer\u003dtp-patch-trunk"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_2c0013aa","line":481,"in_reply_to":"9abb7d3a_b8a446b9","updated":"2016-05-30 16:16:14.000000000","message":"we get the events for all the bridge on the host already, they are currently filtered to get only those that affect br-int","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"e34bc7e7d197abfa5b064c1483f859f71d0566f8","unresolved":false,"context_lines":[{"line_number":478,"context_line":"br-trunk. The OVS agent will be monitoring all the trunk bridges on the host"},{"line_number":479,"context_line":"(currently it monitors only br-int and the ancillary bridges). When it detects"},{"line_number":480,"context_line":"that a new interface has been plugged into a trunk bridge, it will do the"},{"line_number":481,"context_line":"following:"},{"line_number":482,"context_line":" "},{"line_number":483,"context_line":"ovs-vsctl add-port br-trunk tp-patch-trunk -- set Interface tp-patch-trunk type\u003dpatch options:peer\u003dtp-patch-int"},{"line_number":484,"context_line":"ovs-vsctl add-port br-int tp-patch-int tag\u003d3 -- set Interface tp-patch-int type\u003dpatch options:peer\u003dtp-patch-trunk"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_4df82ce6","line":481,"in_reply_to":"9abb7d3a_d7902bef","updated":"2016-05-27 22:54:37.000000000","message":"\u003e I think it is worth noting that bridge deletion should be probably taken care of during trunk deletion as my understanding is that Nova won\u0027t delete the port when a VM boots by port uuid, and thus it only unplug the vif.\n\nNova might not delete the port from the DB, but Nova\u0027s \"unplug\" code at least in the ovs_hybrid case actually deletes the qbr bridge and qvb/qvo veth pair.\n\n\u003e the other reason i prefer agent creating the patch ports is that it will have to create them if subports are added after the vm boots and the agent is the only thing that know the state of the use_veth_interconnection config option\n\nOption C basically requires either the noop firewall driver or an ovs-native firewall driver. Assuming most users still want security groups, they will require the ovs-native firewall driver. Isn\u0027t a high enough version of OVS required for an ovs-native firewall driver that we can basically assume they don\u0027t need to be using \"use_veth_interconnection\"?\n\nIf we don\u0027t have Nova/os-vif create the 1st patch port pair, then the OVS agent now has to monitor more bridges on the host rather than just br-int, all because we wish to maintain compatibility with hosts that can\u0027t support patch port pairs in their old version of OVS.","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"c483fbdf68a2cd69e73b1e501d68e75c1e47212a","unresolved":false,"context_lines":[{"line_number":478,"context_line":"br-trunk. The OVS agent will be monitoring all the trunk bridges on the host"},{"line_number":479,"context_line":"(currently it monitors only br-int and the ancillary bridges). When it detects"},{"line_number":480,"context_line":"that a new interface has been plugged into a trunk bridge, it will do the"},{"line_number":481,"context_line":"following:"},{"line_number":482,"context_line":" "},{"line_number":483,"context_line":"ovs-vsctl add-port br-trunk tp-patch-trunk -- set Interface tp-patch-trunk type\u003dpatch options:peer\u003dtp-patch-int"},{"line_number":484,"context_line":"ovs-vsctl add-port br-int tp-patch-int tag\u003d3 -- set Interface tp-patch-int type\u003dpatch options:peer\u003dtp-patch-trunk"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_d7902bef","line":481,"in_reply_to":"bab6814e_1eecf5b9","updated":"2016-05-27 21:49:40.000000000","message":"I tend to lean on Sean\u0027s suggestion. I think it is worth noting that bridge deletion should be probably taken care of during trunk deletion as my understanding is that Nova won\u0027t delete the port when a VM boots by port uuid, and thus it only unplug the vif.\n\nWe need to sit down and figure out how to integrate this into os-vif asap to minimize effects on Nova, but that\u0027s probably something for yet another discussion.","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"b0d9a56037faf99c1eced82bb7c255f513811f64","unresolved":false,"context_lines":[{"line_number":475,"context_line":"+++++++++++++++++++"},{"line_number":476,"context_line":"A VM is spawn passing to Nova the port-id of a parent port associated with"},{"line_number":477,"context_line":"a trunk port. Nova will create the tap interface and will plug it into"},{"line_number":478,"context_line":"br-trunk. The OVS agent will be monitoring all the trunk bridges on the host"},{"line_number":479,"context_line":"(currently it monitors only br-int and the ancillary bridges). When it detects"},{"line_number":480,"context_line":"that a new interface has been plugged into a trunk bridge, it will do the"},{"line_number":481,"context_line":"following:"},{"line_number":482,"context_line":" "},{"line_number":483,"context_line":"ovs-vsctl add-port br-trunk tp-patch-trunk -- set Interface tp-patch-trunk type\u003dpatch options:peer\u003dtp-patch-int"},{"line_number":484,"context_line":"ovs-vsctl add-port br-int tp-patch-int tag\u003d3 -- set Interface tp-patch-int type\u003dpatch options:peer\u003dtp-patch-trunk"}],"source_content_type":"text/x-rst","patch_set":2,"id":"bab6814e_1eecf5b9","line":481,"range":{"start_line":478,"start_character":10,"end_line":481,"end_character":10},"in_reply_to":"bab6814e_ee69958d","updated":"2016-05-26 22:25:43.000000000","message":"so my input on this is that i would prefer the following\n\nneutron always specifies the bridge to use as part of the vif details.\n\n    The os-vif driver creates the bridge if it does not exist in plug() with the port uuid in the external-ids feild.\n    The os-vif driver add the port to the bdiges as normal setting the external ids feild on the port too.\n    The os-vif driver deletes port only in unplug().\n    The agent plugs the patch cables for all subports and the parent port.\n    The agent monitors the br-int and all bridges with a port id in the external-ids field. if a trunk port is deleted the agent will check if there is a bridge with the same id and delete it and all its patch ports.\n\nso i agree that os-vif/nova should create the bridge. the reason i dont want to have nova/os-vif create the patch port is so we can deprecate and then remove https://github.com/openstack/nova/blob/master/nova/conf/neutron.py#L32-L35.\n\nalternitivly neutron could provide both bridge names in the vif_details in your workflow in which case the os-vif driver\nwould create and delete the trunk port bridge as you suggest and create teh patch port for the trunk port.\n\nin any case i think the patch port tagging should done entirly by the ovs agent the port external-ids should be set on the br-int side by the process that created them\n\nthe other reason i prefer agent creating the patch ports is that it will have to create them if subports are added after the vm boots and the agent is the only thing that know the state of the use_veth_interconnection config option\nhttps://github.com/openstack/neutron/blob/master/neutron/plugins/ml2/drivers/openvswitch/agent/common/config.py#L65-L69","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"c483fbdf68a2cd69e73b1e501d68e75c1e47212a","unresolved":false,"context_lines":[{"line_number":479,"context_line":"(currently it monitors only br-int and the ancillary bridges). When it detects"},{"line_number":480,"context_line":"that a new interface has been plugged into a trunk bridge, it will do the"},{"line_number":481,"context_line":"following:"},{"line_number":482,"context_line":" "},{"line_number":483,"context_line":"ovs-vsctl add-port br-trunk tp-patch-trunk -- set Interface tp-patch-trunk type\u003dpatch options:peer\u003dtp-patch-int"},{"line_number":484,"context_line":"ovs-vsctl add-port br-int tp-patch-int tag\u003d3 -- set Interface tp-patch-int type\u003dpatch options:peer\u003dtp-patch-trunk"},{"line_number":485,"context_line":""}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_17f3d374","line":482,"updated":"2016-05-27 21:49:40.000000000","message":"OMG a space! Shame on you!","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"cca7c99278c322e702f58efd58d0fae4ab1c6a72","unresolved":false,"context_lines":[{"line_number":479,"context_line":"(currently it monitors only br-int and the ancillary bridges). When it detects"},{"line_number":480,"context_line":"that a new interface has been plugged into a trunk bridge, it will do the"},{"line_number":481,"context_line":"following:"},{"line_number":482,"context_line":" "},{"line_number":483,"context_line":"ovs-vsctl add-port br-trunk tp-patch-trunk -- set Interface tp-patch-trunk type\u003dpatch options:peer\u003dtp-patch-int"},{"line_number":484,"context_line":"ovs-vsctl add-port br-int tp-patch-int tag\u003d3 -- set Interface tp-patch-int type\u003dpatch options:peer\u003dtp-patch-trunk"},{"line_number":485,"context_line":""}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_8332b20e","line":482,"in_reply_to":"9abb7d3a_17f3d374","updated":"2016-05-30 16:11:37.000000000","message":"+1 because this made me laugh","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"40dc92ac0d9175b005fe32e57cbfb7f29954eaa0","unresolved":false,"context_lines":[{"line_number":479,"context_line":"(currently it monitors only br-int and the ancillary bridges). When it detects"},{"line_number":480,"context_line":"that a new interface has been plugged into a trunk bridge, it will do the"},{"line_number":481,"context_line":"following:"},{"line_number":482,"context_line":" "},{"line_number":483,"context_line":"ovs-vsctl add-port br-trunk tp-patch-trunk -- set Interface tp-patch-trunk type\u003dpatch options:peer\u003dtp-patch-int"},{"line_number":484,"context_line":"ovs-vsctl add-port br-int tp-patch-int tag\u003d3 -- set Interface tp-patch-int type\u003dpatch options:peer\u003dtp-patch-trunk"},{"line_number":485,"context_line":""}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_a9eb55af","line":482,"in_reply_to":"9abb7d3a_17f3d374","updated":"2016-05-30 16:16:14.000000000","message":":O","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":7776,"name":"Paul Carver","email":"pcarver@att.com","username":"pcarver"},"change_message_id":"d5bb5f553faa37df42be5da9e7cff4ef31b86aab","unresolved":false,"context_lines":[{"line_number":500,"context_line":"ovs-vsctl add-port br-trunk sp-patch-trunk tag\u003d100 -- set Interface sp-patch-trunk type\u003dpatch options:peer\u003dsp-patch-int"},{"line_number":501,"context_line":"ovs-vsctl add-port br-int sp-patch-int tag\u003d5 -- set Interface sp-patch-int type\u003dpatch options:peer\u003dsp-patch-trunk"},{"line_number":502,"context_line":""},{"line_number":503,"context_line":"This patch port connectc the trunk bridge to the integration bridge."},{"line_number":504,"context_line":"sp-patch-trunk, the trunk bridge side of the patch is tagged using vlan 100."},{"line_number":505,"context_line":"We assume that the segmentation ID of the subport is 100."},{"line_number":506,"context_line":"sp-patch-int, the br-int side of the patch port is tagged with vlan 5. We"}],"source_content_type":"text/x-rst","patch_set":2,"id":"bab6814e_857b94bd","line":503,"range":{"start_line":503,"start_character":16,"end_line":503,"end_character":24},"updated":"2016-05-26 12:26:03.000000000","message":"connectc -\u003e connects","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"40dc92ac0d9175b005fe32e57cbfb7f29954eaa0","unresolved":false,"context_lines":[{"line_number":500,"context_line":"ovs-vsctl add-port br-trunk sp-patch-trunk tag\u003d100 -- set Interface sp-patch-trunk type\u003dpatch options:peer\u003dsp-patch-int"},{"line_number":501,"context_line":"ovs-vsctl add-port br-int sp-patch-int tag\u003d5 -- set Interface sp-patch-int type\u003dpatch options:peer\u003dsp-patch-trunk"},{"line_number":502,"context_line":""},{"line_number":503,"context_line":"This patch port connectc the trunk bridge to the integration bridge."},{"line_number":504,"context_line":"sp-patch-trunk, the trunk bridge side of the patch is tagged using vlan 100."},{"line_number":505,"context_line":"We assume that the segmentation ID of the subport is 100."},{"line_number":506,"context_line":"sp-patch-int, the br-int side of the patch port is tagged with vlan 5. We"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9abb7d3a_a902356d","line":503,"range":{"start_line":503,"start_character":16,"end_line":503,"end_character":24},"in_reply_to":"bab6814e_857b94bd","updated":"2016-05-30 16:16:14.000000000","message":"Done","commit_id":"1bcec7f89237a4efc72e31f0085920d9ec47f3bd"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"70c09c8f230393fc6e13fcfd68668c533505e931","unresolved":false,"context_lines":[{"line_number":78,"context_line":""},{"line_number":79,"context_line":"All VM VIFs are plugged into the integration bridge. VM VIFs on a given"},{"line_number":80,"context_line":"virtual network share a common \"local\" VLAN (i.e. not propagated"},{"line_number":81,"context_line":"externally). The VLAN ID of this local VLAN is mapped to the physical"},{"line_number":82,"context_line":"networking details realizing that virtual network."},{"line_number":83,"context_line":""},{"line_number":84,"context_line":"For virtual networks realized as VXLAN/GRE tunnels, a Logical Switch"}],"source_content_type":"text/x-rst","patch_set":3,"id":"9abb7d3a_8abf1fa9","line":81,"updated":"2016-05-31 19:28:59.000000000","message":"this seems extraneous","commit_id":"eb3f2cb07321b6e384a93d360a688546c7e90c7b"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"70c09c8f230393fc6e13fcfd68668c533505e931","unresolved":false,"context_lines":[{"line_number":303,"context_line":""},{"line_number":304,"context_line":"Implementation VLAN Interfaces (Option A)"},{"line_number":305,"context_line":"~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~"},{"line_number":306,"context_line":"This implementation doesn\u0027t require any modification of the vif-drivers since"},{"line_number":307,"context_line":"Nova will plug the vif of the VM the same way as it does for traditional ports."},{"line_number":308,"context_line":""},{"line_number":309,"context_line":"Trunk port creation"}],"source_content_type":"text/x-rst","patch_set":3,"id":"9abb7d3a_aa1e63b8","line":306,"range":{"start_line":306,"start_character":32,"end_line":306,"end_character":33},"updated":"2016-05-31 19:28:59.000000000","message":"space?","commit_id":"eb3f2cb07321b6e384a93d360a688546c7e90c7b"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"70c09c8f230393fc6e13fcfd68668c533505e931","unresolved":false,"context_lines":[{"line_number":321,"context_line":"server that the trunk port is up. Neutron will send back to Nova an event to"},{"line_number":322,"context_line":"signal that the wiring was successful."},{"line_number":323,"context_line":"If the parent port is associated with one or more subports the agent will process"},{"line_number":324,"context_line":"then as described in the next paragraph."},{"line_number":325,"context_line":""},{"line_number":326,"context_line":"Subport creation"},{"line_number":327,"context_line":"++++++++++++++++"}],"source_content_type":"text/x-rst","patch_set":3,"id":"9abb7d3a_50016637","line":324,"updated":"2016-05-31 19:28:59.000000000","message":"then -\u003e them?","commit_id":"eb3f2cb07321b6e384a93d360a688546c7e90c7b"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"70c09c8f230393fc6e13fcfd68668c533505e931","unresolved":false,"context_lines":[{"line_number":323,"context_line":"If the parent port is associated with one or more subports the agent will process"},{"line_number":324,"context_line":"then as described in the next paragraph."},{"line_number":325,"context_line":""},{"line_number":326,"context_line":"Subport creation"},{"line_number":327,"context_line":"++++++++++++++++"},{"line_number":328,"context_line":"If a subport is added to a parent port but no VM was booted using that parent port"},{"line_number":329,"context_line":"yet, the agent won\u0027t process the subport."}],"source_content_type":"text/x-rst","patch_set":3,"id":"9abb7d3a_b001b2b1","line":326,"updated":"2016-05-31 19:28:59.000000000","message":"I wonder if we can collapse subport creation/deletion scenarios for conciseness, but not a big issue.","commit_id":"eb3f2cb07321b6e384a93d360a688546c7e90c7b"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"70c09c8f230393fc6e13fcfd68668c533505e931","unresolved":false,"context_lines":[{"line_number":326,"context_line":"Subport creation"},{"line_number":327,"context_line":"++++++++++++++++"},{"line_number":328,"context_line":"If a subport is added to a parent port but no VM was booted using that parent port"},{"line_number":329,"context_line":"yet, the agent won\u0027t process the subport."},{"line_number":330,"context_line":"When a subport is created for a parent port and a VM that uses that parent port is"},{"line_number":331,"context_line":"already running, the OVS agent will create a VLAN interface on the VM tap"},{"line_number":332,"context_line":"using the VLAN ID specified in the subport segmentation id."}],"source_content_type":"text/x-rst","patch_set":3,"id":"9abb7d3a_70746a67","line":329,"updated":"2016-05-31 19:28:59.000000000","message":"I\u0027d add (because at this point there\u0027s no node associated with the parent port).","commit_id":"eb3f2cb07321b6e384a93d360a688546c7e90c7b"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"32916609bc5ca07195fc09251b48db472a188f0b","unresolved":false,"context_lines":[{"line_number":411,"context_line":"*Inbound traffic from the VM point of view*"},{"line_number":412,"context_line":""},{"line_number":413,"context_line":"The traffic will flow untagged from port 1 to eth0. The traffic going out from port 2"},{"line_number":414,"context_line":"will be stripped of the VLAN ID assigned to network1. It will be filtered by the rules"},{"line_number":415,"context_line":"installed by the firewall and reach tap1.100."},{"line_number":416,"context_line":"tap1.100 will tag the traffic using VLAN 100. It will then reach the VM\u0027s eth0.100."},{"line_number":417,"context_line":""}],"source_content_type":"text/x-rst","patch_set":3,"id":"9abb7d3a_87ab9574","line":414,"range":{"start_line":414,"start_character":51,"end_line":414,"end_character":52},"updated":"2016-05-31 20:29:58.000000000","message":"2","commit_id":"eb3f2cb07321b6e384a93d360a688546c7e90c7b"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"70c09c8f230393fc6e13fcfd68668c533505e931","unresolved":false,"context_lines":[{"line_number":427,"context_line":"Trunk port deletion"},{"line_number":428,"context_line":"+++++++++++++++++++"},{"line_number":429,"context_line":""},{"line_number":430,"context_line":"Deleting a port that is an active parent in a trunk is forbidden. When Nova deletes a VM,"},{"line_number":431,"context_line":"it ends up deleting the associated VIFs if it has created the port, but since the port"},{"line_number":432,"context_line":"will be passed to Nova, Nova won\u0027t touch it. The OVS agent should take care of it."},{"line_number":433,"context_line":"When a trunk port is deleted, all the subports associated with that trunk should be"}],"source_content_type":"text/x-rst","patch_set":3,"id":"9abb7d3a_c4eabb0e","line":430,"updated":"2016-05-31 19:28:59.000000000","message":"To be perfectly clear: today a Neutron user can go and delete a port associated to a VM and Nova gets notified. Now we\u0027re saying that if a port is engaged in a trunk, the operation is prevented. This is reasonable to me. However, a user can go and issue a trunk-delete. In this case this should still result in the same vif-unplug notification.\n\nSo is this section about trunk deletion or deletion of the parent port? Maybe more clarity would be beneficial to the reader.","commit_id":"eb3f2cb07321b6e384a93d360a688546c7e90c7b"},{"author":{"_account_id":4187,"name":"Ryan Tidwell","email":"rtidwell@suse.com","username":"ryan-tidwell"},"change_message_id":"6f3e81714054b660bef2ec5d18168b839329ceba","unresolved":false,"context_lines":[{"line_number":427,"context_line":"Trunk port deletion"},{"line_number":428,"context_line":"+++++++++++++++++++"},{"line_number":429,"context_line":""},{"line_number":430,"context_line":"Deleting a port that is an active parent in a trunk is forbidden. When Nova deletes a VM,"},{"line_number":431,"context_line":"it ends up deleting the associated VIFs if it has created the port, but since the port"},{"line_number":432,"context_line":"will be passed to Nova, Nova won\u0027t touch it. The OVS agent should take care of it."},{"line_number":433,"context_line":"When a trunk port is deleted, all the subports associated with that trunk should be"}],"source_content_type":"text/x-rst","patch_set":3,"id":"7aa08908_93393ef5","line":430,"in_reply_to":"9abb7d3a_c4eabb0e","updated":"2016-06-07 16:44:22.000000000","message":"Does the removal of a parent port really need to be prevented?  Removal of the parent port would result in the cascading removal of the associated trunk and subports. Is this really something we should prevent? It actually seems ok to me, and is what I would expect. I don\u0027t think we need to change any of conditions related to port delete, I think it should work fine as is.","commit_id":"eb3f2cb07321b6e384a93d360a688546c7e90c7b"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"70c09c8f230393fc6e13fcfd68668c533505e931","unresolved":false,"context_lines":[{"line_number":429,"context_line":""},{"line_number":430,"context_line":"Deleting a port that is an active parent in a trunk is forbidden. When Nova deletes a VM,"},{"line_number":431,"context_line":"it ends up deleting the associated VIFs if it has created the port, but since the port"},{"line_number":432,"context_line":"will be passed to Nova, Nova won\u0027t touch it. The OVS agent should take care of it."},{"line_number":433,"context_line":"When a trunk port is deleted, all the subports associated with that trunk should be"},{"line_number":434,"context_line":"deleted too. To recognize them, the trunk port ID might be stored in the subport"},{"line_number":435,"context_line":"external_ids."}],"source_content_type":"text/x-rst","patch_set":3,"id":"9abb7d3a_3097823b","line":432,"updated":"2016-05-31 19:28:59.000000000","message":"This last sentence is a bit ambiguous. Take care of what exactly?","commit_id":"eb3f2cb07321b6e384a93d360a688546c7e90c7b"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"70c09c8f230393fc6e13fcfd68668c533505e931","unresolved":false,"context_lines":[{"line_number":430,"context_line":"Deleting a port that is an active parent in a trunk is forbidden. When Nova deletes a VM,"},{"line_number":431,"context_line":"it ends up deleting the associated VIFs if it has created the port, but since the port"},{"line_number":432,"context_line":"will be passed to Nova, Nova won\u0027t touch it. The OVS agent should take care of it."},{"line_number":433,"context_line":"When a trunk port is deleted, all the subports associated with that trunk should be"},{"line_number":434,"context_line":"deleted too. To recognize them, the trunk port ID might be stored in the subport"},{"line_number":435,"context_line":"external_ids."},{"line_number":436,"context_line":""}],"source_content_type":"text/x-rst","patch_set":3,"id":"9abb7d3a_50a2069a","line":433,"updated":"2016-05-31 19:28:59.000000000","message":"This should be: \u0027when a trunk is deleted\u0027.","commit_id":"eb3f2cb07321b6e384a93d360a688546c7e90c7b"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"70c09c8f230393fc6e13fcfd68668c533505e931","unresolved":false,"context_lines":[{"line_number":442,"context_line":"the OVS agent will take care of removing the VLAN interface, the firewall bridge"},{"line_number":443,"context_line":"if using iptables firewall and the port on br-int."},{"line_number":444,"context_line":""},{"line_number":445,"context_line":"Agent resync"},{"line_number":446,"context_line":"++++++++++++"},{"line_number":447,"context_line":""},{"line_number":448,"context_line":"During resync the agent should check that all the trunk and subports on br-int are still valid and"}],"source_content_type":"text/x-rst","patch_set":3,"id":"9abb7d3a_f0de9a0f","line":445,"updated":"2016-05-31 19:28:59.000000000","message":"should this be a section on its own? I don\u0027t see much difference between the two options.","commit_id":"eb3f2cb07321b6e384a93d360a688546c7e90c7b"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"4eb3e99b841e7553741391157de02767ec819117","unresolved":false,"context_lines":[{"line_number":445,"context_line":"Agent resync"},{"line_number":446,"context_line":"++++++++++++"},{"line_number":447,"context_line":""},{"line_number":448,"context_line":"During resync the agent should check that all the trunk and subports on br-int are still valid and"},{"line_number":449,"context_line":"delete the ones that are stale."},{"line_number":450,"context_line":""},{"line_number":451,"context_line":"Implementation Trunk Bridge (Option C)"},{"line_number":452,"context_line":"~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~"}],"source_content_type":"text/x-rst","patch_set":3,"id":"9abb7d3a_5bafb029","line":449,"range":{"start_line":448,"start_character":0,"end_line":449,"end_character":31},"updated":"2016-06-01 09:28:41.000000000","message":"Plus the agent should delete leftover vlan@tap interfaces too if there are any.","commit_id":"eb3f2cb07321b6e384a93d360a688546c7e90c7b"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"32916609bc5ca07195fc09251b48db472a188f0b","unresolved":false,"context_lines":[{"line_number":452,"context_line":"~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~"},{"line_number":453,"context_line":"This implementation is based on this `_etherpad \u003chttps://etherpad.openstack.org/p/trunk-bridge-tagged-patch-experiment\u003e`."},{"line_number":454,"context_line":"Credits to Bence Romsics."},{"line_number":455,"context_line":"If use_veth_interconnection\u003dtrue instead of using patch port, veth pairs will be used."},{"line_number":456,"context_line":""},{"line_number":457,"context_line":"::"},{"line_number":458,"context_line":""}],"source_content_type":"text/x-rst","patch_set":3,"id":"9abb7d3a_87f0b592","line":455,"range":{"start_line":455,"start_character":0,"end_line":455,"end_character":86},"updated":"2016-05-31 20:29:58.000000000","message":"This might not be necessary, see https://bugs.launchpad.net/neutron/+bug/1550501/comments/1","commit_id":"eb3f2cb07321b6e384a93d360a688546c7e90c7b"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"0da831444f0162d74c70992dcc807798237b2308","unresolved":false,"context_lines":[{"line_number":452,"context_line":"~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~"},{"line_number":453,"context_line":"This implementation is based on this `_etherpad \u003chttps://etherpad.openstack.org/p/trunk-bridge-tagged-patch-experiment\u003e`."},{"line_number":454,"context_line":"Credits to Bence Romsics."},{"line_number":455,"context_line":"If use_veth_interconnection\u003dtrue instead of using patch port, veth pairs will be used."},{"line_number":456,"context_line":""},{"line_number":457,"context_line":"::"},{"line_number":458,"context_line":""}],"source_content_type":"text/x-rst","patch_set":3,"id":"9abb7d3a_214f2c6c","line":455,"range":{"start_line":455,"start_character":0,"end_line":455,"end_character":86},"in_reply_to":"9abb7d3a_87f0b592","updated":"2016-06-01 02:03:38.000000000","message":"i think suse also required this option but i dont know if there is a supported suse release that runs a supported openstack that does not support patch ports.\n\nits really and edge case of an edge case but its easy to support and simple to remove when use_veth_interconnection is removed.","commit_id":"eb3f2cb07321b6e384a93d360a688546c7e90c7b"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"32916609bc5ca07195fc09251b48db472a188f0b","unresolved":false,"context_lines":[{"line_number":488,"context_line":"A VM is spawned passing to Nova the port-id of a parent port associated with"},{"line_number":489,"context_line":"a trunk port and the bridge to use as part of the vif details."},{"line_number":490,"context_line":"The os-vif driver creates the trunk bridge br-trunk-1 if it does not exist in plug()"},{"line_number":491,"context_line":"with the parent port ID in the external-ids field. It will create the tap interface"},{"line_number":492,"context_line":"tap1 and plug it into br-trunk-1 setting the parent port ID in the external-ids."},{"line_number":493,"context_line":"The OVS agent will be monitoring all the trunk bridges on the host"},{"line_number":494,"context_line":"(currently it monitors only br-int and the ancillary bridges). When it detects"}],"source_content_type":"text/x-rst","patch_set":3,"id":"9abb7d3a_5a534039","line":491,"range":{"start_line":491,"start_character":0,"end_line":491,"end_character":49},"updated":"2016-05-31 20:29:58.000000000","message":"This seems redundant. Do we need the parent port ID in both the external-ids field of the bridge *and* the port for the tap device?","commit_id":"eb3f2cb07321b6e384a93d360a688546c7e90c7b"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"0da831444f0162d74c70992dcc807798237b2308","unresolved":false,"context_lines":[{"line_number":488,"context_line":"A VM is spawned passing to Nova the port-id of a parent port associated with"},{"line_number":489,"context_line":"a trunk port and the bridge to use as part of the vif details."},{"line_number":490,"context_line":"The os-vif driver creates the trunk bridge br-trunk-1 if it does not exist in plug()"},{"line_number":491,"context_line":"with the parent port ID in the external-ids field. It will create the tap interface"},{"line_number":492,"context_line":"tap1 and plug it into br-trunk-1 setting the parent port ID in the external-ids."},{"line_number":493,"context_line":"The OVS agent will be monitoring all the trunk bridges on the host"},{"line_number":494,"context_line":"(currently it monitors only br-int and the ancillary bridges). When it detects"}],"source_content_type":"text/x-rst","patch_set":3,"id":"9abb7d3a_b60ec038","line":491,"range":{"start_line":491,"start_character":0,"end_line":491,"end_character":49},"in_reply_to":"9abb7d3a_5a534039","updated":"2016-06-01 02:03:38.000000000","message":"not strictly but the idea was that the agent could monitor port deletions whcih would be done by nova/os-vif and then find the bridge with the same id and clean up the patch ports and delete the bridge.","commit_id":"eb3f2cb07321b6e384a93d360a688546c7e90c7b"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"70c09c8f230393fc6e13fcfd68668c533505e931","unresolved":false,"context_lines":[{"line_number":490,"context_line":"The os-vif driver creates the trunk bridge br-trunk-1 if it does not exist in plug()"},{"line_number":491,"context_line":"with the parent port ID in the external-ids field. It will create the tap interface"},{"line_number":492,"context_line":"tap1 and plug it into br-trunk-1 setting the parent port ID in the external-ids."},{"line_number":493,"context_line":"The OVS agent will be monitoring all the trunk bridges on the host"},{"line_number":494,"context_line":"(currently it monitors only br-int and the ancillary bridges). When it detects"},{"line_number":495,"context_line":"that a new interface has been plugged into a trunk bridge, it will do the"},{"line_number":496,"context_line":"following:"}],"source_content_type":"text/x-rst","patch_set":3,"id":"9abb7d3a_5053266b","line":493,"updated":"2016-05-31 19:28:59.000000000","message":"so it will have to monitor the creation of new OVS bridges.","commit_id":"eb3f2cb07321b6e384a93d360a688546c7e90c7b"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"70c09c8f230393fc6e13fcfd68668c533505e931","unresolved":false,"context_lines":[{"line_number":504,"context_line":"tag. It will carry untagged traffic."},{"line_number":505,"context_line":"tp-patch-int, the br-int side the patch port is tagged with VLAN 3. We assume that the"},{"line_number":506,"context_line":"trunk port is on network1 that on this host is associated with VLAN 3."},{"line_number":507,"context_line":"The OVS agent will set the trunk port ID in the external-ids of tp-patch-trunk and tp-patch-int."},{"line_number":508,"context_line":""},{"line_number":509,"context_line":"Subport creation"},{"line_number":510,"context_line":"++++++++++++++++"}],"source_content_type":"text/x-rst","patch_set":3,"id":"9abb7d3a_30e662b0","line":507,"updated":"2016-05-31 19:28:59.000000000","message":"I\u0027d add:\n\nIf the parent port is associated with one or more subports the agent will process them as described in the next paragraph.","commit_id":"eb3f2cb07321b6e384a93d360a688546c7e90c7b"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"4eb3e99b841e7553741391157de02767ec819117","unresolved":false,"context_lines":[{"line_number":504,"context_line":"tag. It will carry untagged traffic."},{"line_number":505,"context_line":"tp-patch-int, the br-int side the patch port is tagged with VLAN 3. We assume that the"},{"line_number":506,"context_line":"trunk port is on network1 that on this host is associated with VLAN 3."},{"line_number":507,"context_line":"The OVS agent will set the trunk port ID in the external-ids of tp-patch-trunk and tp-patch-int."},{"line_number":508,"context_line":""},{"line_number":509,"context_line":"Subport creation"},{"line_number":510,"context_line":"++++++++++++++++"}],"source_content_type":"text/x-rst","patch_set":3,"id":"9abb7d3a_57cdc42f","line":507,"in_reply_to":"9abb7d3a_30e662b0","updated":"2016-06-01 09:28:41.000000000","message":"Do we want to label by external_ids *all* ovs objects belonging to a neutron port? It would make sense, for example could simplify cleanup  methods (\"Delete all ovs ports and bridges with a particular external_id.\" - This could work for regular and trunk ports too.)\n\nIf we want this, we should probably state it.","commit_id":"eb3f2cb07321b6e384a93d360a688546c7e90c7b"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"70c09c8f230393fc6e13fcfd68668c533505e931","unresolved":false,"context_lines":[{"line_number":509,"context_line":"Subport creation"},{"line_number":510,"context_line":"++++++++++++++++"},{"line_number":511,"context_line":""},{"line_number":512,"context_line":"When a subport is created the OVS agent will create a new patch port:"},{"line_number":513,"context_line":""},{"line_number":514,"context_line":""},{"line_number":515,"context_line":"ovs-vsctl add-port  br-trunk-1 sp-patch-trunk tag\u003d100 -- set Interface sp-patch-trunk type\u003dpatch options:peer\u003dsp-patch-int"}],"source_content_type":"text/x-rst","patch_set":3,"id":"9abb7d3a_f0197a89","line":512,"updated":"2016-05-31 19:28:59.000000000","message":"this happens only if the trunk is bound to a VM","commit_id":"eb3f2cb07321b6e384a93d360a688546c7e90c7b"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"0da831444f0162d74c70992dcc807798237b2308","unresolved":false,"context_lines":[{"line_number":509,"context_line":"Subport creation"},{"line_number":510,"context_line":"++++++++++++++++"},{"line_number":511,"context_line":""},{"line_number":512,"context_line":"When a subport is created the OVS agent will create a new patch port:"},{"line_number":513,"context_line":""},{"line_number":514,"context_line":""},{"line_number":515,"context_line":"ovs-vsctl add-port  br-trunk-1 sp-patch-trunk tag\u003d100 -- set Interface sp-patch-trunk type\u003dpatch options:peer\u003dsp-patch-int"}],"source_content_type":"text/x-rst","patch_set":3,"id":"9abb7d3a_166a2cee","line":512,"in_reply_to":"9abb7d3a_f0197a89","updated":"2016-06-01 02:03:38.000000000","message":"that is true but to qualify your clarification \nyou were refering to timing not respociblity for patch\nport creation.\ne.g.\nyou were saying that the ovs agent will only create the patch port after the vm interface has been plugged into the ovs controlled by the agent \n\nas distinct form \n\nit will not be the responsible of the ovs agent to create the patch port if the trunk was not bound to a vm when the subport was created.","commit_id":"eb3f2cb07321b6e384a93d360a688546c7e90c7b"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"70c09c8f230393fc6e13fcfd68668c533505e931","unresolved":false,"context_lines":[{"line_number":542,"context_line":"Trunk port deletion"},{"line_number":543,"context_line":"+++++++++++++++++++"},{"line_number":544,"context_line":""},{"line_number":545,"context_line":"The OVS agent will delete the trunk bridge corresponding to the trunk port."},{"line_number":546,"context_line":""},{"line_number":547,"context_line":"Subport deletion"},{"line_number":548,"context_line":"+++++++++++++++++++"}],"source_content_type":"text/x-rst","patch_set":3,"id":"9abb7d3a_2412a757","line":545,"updated":"2016-05-31 19:28:59.000000000","message":"is the trunk bridge deletion (and subsequent subports deletion) a consequence of the VIF unplugging? See my comment above. It\u0027s a bit unclear to me whether this is an aftermath of VM delete or trunk-delete. It\u0027s maybe the former, but I\u0027d leave nothing to doubt.","commit_id":"eb3f2cb07321b6e384a93d360a688546c7e90c7b"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"0da831444f0162d74c70992dcc807798237b2308","unresolved":false,"context_lines":[{"line_number":542,"context_line":"Trunk port deletion"},{"line_number":543,"context_line":"+++++++++++++++++++"},{"line_number":544,"context_line":""},{"line_number":545,"context_line":"The OVS agent will delete the trunk bridge corresponding to the trunk port."},{"line_number":546,"context_line":""},{"line_number":547,"context_line":"Subport deletion"},{"line_number":548,"context_line":"+++++++++++++++++++"}],"source_content_type":"text/x-rst","patch_set":3,"id":"9abb7d3a_d6d6e4ef","line":545,"in_reply_to":"9abb7d3a_2412a757","updated":"2016-06-01 02:03:38.000000000","message":"i think the trunk bridge should be deleted by the ovs agent when nova/os-vif unplugs the vm interface coresponding to the trunk port from the bridge.\n\ni think this workflow is required not to have bridge stick around if i was to live migrate the instance form one host to another. in this case the trunk will not be deleted in the neutron database but the host id will be updated.","commit_id":"eb3f2cb07321b6e384a93d360a688546c7e90c7b"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"4eb3e99b841e7553741391157de02767ec819117","unresolved":false,"context_lines":[{"line_number":542,"context_line":"Trunk port deletion"},{"line_number":543,"context_line":"+++++++++++++++++++"},{"line_number":544,"context_line":""},{"line_number":545,"context_line":"The OVS agent will delete the trunk bridge corresponding to the trunk port."},{"line_number":546,"context_line":""},{"line_number":547,"context_line":"Subport deletion"},{"line_number":548,"context_line":"+++++++++++++++++++"}],"source_content_type":"text/x-rst","patch_set":3,"id":"9abb7d3a_9aad919d","line":545,"in_reply_to":"9abb7d3a_d6d6e4ef","updated":"2016-06-01 09:28:41.000000000","message":"+1 to trigger cleanup of trunk bridges by unplug().\n\nI like symmetry in setup and teardown. That is if trunk bridge (plus patch cables etc) creation is triggered by plug() then it\u0027s easier if trunk bridge deletion (etc) is triggered by unplug().\n\nFor example think of nova shelve/unshelve. Shelve will call unplug(), but will delete neither the parent port, nor the trunk. The trunk bridge should not left around because unshelve will likely go to another host.","commit_id":"eb3f2cb07321b6e384a93d360a688546c7e90c7b"},{"author":{"_account_id":4187,"name":"Ryan Tidwell","email":"rtidwell@suse.com","username":"ryan-tidwell"},"change_message_id":"6f3e81714054b660bef2ec5d18168b839329ceba","unresolved":false,"context_lines":[{"line_number":547,"context_line":"Subport deletion"},{"line_number":548,"context_line":"+++++++++++++++++++"},{"line_number":549,"context_line":""},{"line_number":550,"context_line":"The OVS agent will delete the patch port corresponding to the subport deleted."},{"line_number":551,"context_line":""},{"line_number":552,"context_line":"Agent resync"},{"line_number":553,"context_line":"++++++++++++"}],"source_content_type":"text/x-rst","patch_set":3,"id":"7aa08908_e5743569","line":550,"updated":"2016-06-07 16:44:22.000000000","message":"If the trunk bridge is pulled out from underneath existing patch ports, do those patch ports get deleted? At the model layer removal of a trunk results in a cascading delete of all associated subports. Do we get that same behavior in the data path? That would be a good experiment to run. If we don\u0027t have to explicitly clean up subports in the data path life is awesome!","commit_id":"eb3f2cb07321b6e384a93d360a688546c7e90c7b"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"0da831444f0162d74c70992dcc807798237b2308","unresolved":false,"context_lines":[{"line_number":552,"context_line":"Agent resync"},{"line_number":553,"context_line":"++++++++++++"},{"line_number":554,"context_line":""},{"line_number":555,"context_line":"During resync the agent should check that all the trunk and subports on br-int are"},{"line_number":556,"context_line":"still valid and delete the ones that are stale."},{"line_number":557,"context_line":""},{"line_number":558,"context_line":""},{"line_number":559,"context_line":"Further Reading"}],"source_content_type":"text/x-rst","patch_set":3,"id":"9abb7d3a_b641a0ab","line":556,"range":{"start_line":555,"start_character":0,"end_line":556,"end_character":47},"updated":"2016-06-01 02:03:38.000000000","message":"i think this is a little unclear. \nthe trunk and subports in this model are patch port pairs that exist between the br-int and trunk bridge.","commit_id":"eb3f2cb07321b6e384a93d360a688546c7e90c7b"},{"author":{"_account_id":4187,"name":"Ryan Tidwell","email":"rtidwell@suse.com","username":"ryan-tidwell"},"change_message_id":"b79bd7f71cf9449cb9ca9c5707efe7d394f7c373","unresolved":false,"context_lines":[{"line_number":555,"context_line":"During resync the agent should check that all the trunk and subports on br-int are"},{"line_number":556,"context_line":"still valid and delete the ones that are stale."},{"line_number":557,"context_line":""},{"line_number":558,"context_line":""},{"line_number":559,"context_line":"Further Reading"},{"line_number":560,"context_line":"---------------"},{"line_number":561,"context_line":""}],"source_content_type":"text/x-rst","patch_set":3,"id":"7aa08908_26ee38b4","line":558,"updated":"2016-06-07 21:04:02.000000000","message":"Can we have a section dedicated to VIF plugging. We seem to be waving our hands around it a little bit. There are some serious pros and cons to consider.\n\nIf we let nova plug br-trunk to br-int then monitor for that action, there is information about the trunk that needs to be passed over to nova (ie use veth pair or patch port pair).\n\nIf we just monitor for a new bridge to pop up and have the agent wire between br-trunk and br-int, there\u0027s a lot of code in neutron that assumes it\u0027s dealing with the integration bridge.  It also assumes Nova has plugged the port, so we would end up with another trunk-specific mechanism for wiring the VIF and trunk bridge. If we\u0027re not careful, we could end up dumping a lot of trunk-specific logic into the OVS agent and coupling it pretty tightly.\n\nI think vif plugging warrants its own section so it can be explained and hashed out.","commit_id":"eb3f2cb07321b6e384a93d360a688546c7e90c7b"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"43698aeb7cd95013d0c1a617d0b9c5f1761e5173","unresolved":false,"context_lines":[{"line_number":555,"context_line":"During resync the agent should check that all the trunk and subports on br-int are"},{"line_number":556,"context_line":"still valid and delete the ones that are stale."},{"line_number":557,"context_line":""},{"line_number":558,"context_line":""},{"line_number":559,"context_line":"Further Reading"},{"line_number":560,"context_line":"---------------"},{"line_number":561,"context_line":""}],"source_content_type":"text/x-rst","patch_set":3,"id":"7aa08908_215e2212","line":558,"in_reply_to":"7aa08908_26ee38b4","updated":"2016-06-07 21:14:58.000000000","message":"\u003e Can we have a section dedicated to VIF plugging.\n\n+1 I think we need this as well.\n\n\u003e If we let nova plug br-trunk to br-int then monitor for that action, there is information about the trunk that needs to be passed over to nova (ie use veth pair or patch port pair).\n\nI will also add here that letting Nova plug the 1st patch port pair into br-int allows us to make use of the current L2 agent extension framework as-is, because the extension driver would be triggered from the new port being added to br-int.","commit_id":"eb3f2cb07321b6e384a93d360a688546c7e90c7b"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"70c09c8f230393fc6e13fcfd68668c533505e931","unresolved":false,"context_lines":[{"line_number":559,"context_line":"Further Reading"},{"line_number":560,"context_line":"---------------"},{"line_number":561,"context_line":""},{"line_number":562,"context_line":"* `Darragh O\u0027Reilly - The Open vSwitch plugin with VLANs \u003chttp://techbackground.blogspot.com/2013/07/the-open-vswitch-plugin-with-VLANs.html\u003e`_"}],"source_content_type":"text/x-rst","patch_set":3,"id":"9abb7d3a_64502fd5","line":562,"updated":"2016-05-31 19:28:59.000000000","message":"you were too aggressive with your replace logic. You broke the link","commit_id":"eb3f2cb07321b6e384a93d360a688546c7e90c7b"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"5f2a48a56469f2a042ead7b55cbfe6ff908d9db1","unresolved":false,"context_lines":[{"line_number":309,"context_line":"Trunk port creation"},{"line_number":310,"context_line":"+++++++++++++++++++"},{"line_number":311,"context_line":"A VM is spawned passing to Nova the port-id of a parent port associated with"},{"line_number":312,"context_line":"a trunk port. Nova will create the tap interface and will plug it into"},{"line_number":313,"context_line":"br-int or into the firewall bridge if using iptables firewall. In the"},{"line_number":314,"context_line":"external-ids of the port Nova will store the port ID of the parent port."},{"line_number":315,"context_line":"The OVS agent detects that a new vif has been plugged. It gets"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_a81732a1","line":312,"range":{"start_line":312,"start_character":2,"end_line":312,"end_character":12},"updated":"2016-06-09 10:01:48.000000000","message":"nit: s/trunk port/trunk/","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"5538f5b25cc28c31e568eadea620fd221987b714","unresolved":false,"context_lines":[{"line_number":309,"context_line":"Trunk port creation"},{"line_number":310,"context_line":"+++++++++++++++++++"},{"line_number":311,"context_line":"A VM is spawned passing to Nova the port-id of a parent port associated with"},{"line_number":312,"context_line":"a trunk port. Nova will create the tap interface and will plug it into"},{"line_number":313,"context_line":"br-int or into the firewall bridge if using iptables firewall. In the"},{"line_number":314,"context_line":"external-ids of the port Nova will store the port ID of the parent port."},{"line_number":315,"context_line":"The OVS agent detects that a new vif has been plugged. It gets"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_7c01c904","line":312,"in_reply_to":"7aa08908_9e8fa922","updated":"2016-06-14 13:36:11.000000000","message":"I agree, thanks for the suggestion!","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":7448,"name":"Carl Baldwin","email":"carl@ecbaldwin.net","username":"carl-baldwin"},"change_message_id":"ad5cc759a71392ee4185b91753fe954cf5ab94da","unresolved":false,"context_lines":[{"line_number":309,"context_line":"Trunk port creation"},{"line_number":310,"context_line":"+++++++++++++++++++"},{"line_number":311,"context_line":"A VM is spawned passing to Nova the port-id of a parent port associated with"},{"line_number":312,"context_line":"a trunk port. Nova will create the tap interface and will plug it into"},{"line_number":313,"context_line":"br-int or into the firewall bridge if using iptables firewall. In the"},{"line_number":314,"context_line":"external-ids of the port Nova will store the port ID of the parent port."},{"line_number":315,"context_line":"The OVS agent detects that a new vif has been plugged. It gets"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_9e8fa922","line":312,"in_reply_to":"7aa08908_a81732a1","updated":"2016-06-10 22:10:01.000000000","message":"+1 I think this will read more easily if we\u0027re consistently distinguishing between the \"trunk\" (which is not a port, but has a parent port and subports) and the parent port.  The term \"trunk port\" sounds like it could refer to either.  So, I think we should avoid the term \"trunk port\" altogether and stick with these:\n\n  trunk\n  parent port\n  subport","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"5f2a48a56469f2a042ead7b55cbfe6ff908d9db1","unresolved":false,"context_lines":[{"line_number":316,"context_line":"the details of the new port and wires it."},{"line_number":317,"context_line":"The agent configures it in the same way as a traditional port: packets coming out"},{"line_number":318,"context_line":"from the VM will be tagged using the internal VLAN ID associated to the network,"},{"line_number":319,"context_line":"packets going to the VM will be stripped of the VLAN id."},{"line_number":320,"context_line":"After wiring it successfully the OVS agent will send a message notifying Neutron"},{"line_number":321,"context_line":"server that the trunk port is up. Neutron will send back to Nova an event to"},{"line_number":322,"context_line":"signal that the wiring was successful."}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_881fd6b5","line":319,"range":{"start_line":319,"start_character":53,"end_line":319,"end_character":55},"updated":"2016-06-09 10:01:48.000000000","message":"nit: s/id/ID/","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"5538f5b25cc28c31e568eadea620fd221987b714","unresolved":false,"context_lines":[{"line_number":316,"context_line":"the details of the new port and wires it."},{"line_number":317,"context_line":"The agent configures it in the same way as a traditional port: packets coming out"},{"line_number":318,"context_line":"from the VM will be tagged using the internal VLAN ID associated to the network,"},{"line_number":319,"context_line":"packets going to the VM will be stripped of the VLAN id."},{"line_number":320,"context_line":"After wiring it successfully the OVS agent will send a message notifying Neutron"},{"line_number":321,"context_line":"server that the trunk port is up. Neutron will send back to Nova an event to"},{"line_number":322,"context_line":"signal that the wiring was successful."}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_5c2b2d82","line":319,"range":{"start_line":319,"start_character":53,"end_line":319,"end_character":55},"in_reply_to":"7aa08908_881fd6b5","updated":"2016-06-14 13:36:11.000000000","message":"Done","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":7448,"name":"Carl Baldwin","email":"carl@ecbaldwin.net","username":"carl-baldwin"},"change_message_id":"ad5cc759a71392ee4185b91753fe954cf5ab94da","unresolved":false,"context_lines":[{"line_number":327,"context_line":"++++++++++++++++"},{"line_number":328,"context_line":"If a subport is added to a parent port but no VM was booted using that parent port"},{"line_number":329,"context_line":"yet, the agent won\u0027t process the subport (because at this point there\u0027s no node"},{"line_number":330,"context_line":"associated with the parent port)."},{"line_number":331,"context_line":"When a subport is created for a parent port and a VM that uses that parent port is"},{"line_number":332,"context_line":"already running, the OVS agent will create a VLAN interface on the VM tap"},{"line_number":333,"context_line":"using the VLAN ID specified in the subport segmentation id."}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_19275313","line":330,"updated":"2016-06-10 22:10:01.000000000","message":"It probably isn\u0027t bound yet, either.  Right?  So, which agent would process it even if it wanted to.","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"5538f5b25cc28c31e568eadea620fd221987b714","unresolved":false,"context_lines":[{"line_number":327,"context_line":"++++++++++++++++"},{"line_number":328,"context_line":"If a subport is added to a parent port but no VM was booted using that parent port"},{"line_number":329,"context_line":"yet, the agent won\u0027t process the subport (because at this point there\u0027s no node"},{"line_number":330,"context_line":"associated with the parent port)."},{"line_number":331,"context_line":"When a subport is created for a parent port and a VM that uses that parent port is"},{"line_number":332,"context_line":"already running, the OVS agent will create a VLAN interface on the VM tap"},{"line_number":333,"context_line":"using the VLAN ID specified in the subport segmentation id."}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_5c796d92","line":330,"in_reply_to":"7aa08908_19275313","updated":"2016-06-14 13:36:11.000000000","message":"Right no agent would process it because the parent port is not bound to any host. I will clarify.","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"185dbe523a2e243a01e3aa5cf06a93a9a0efa175","unresolved":false,"context_lines":[{"line_number":399,"context_line":"           +---+---+       +-----+-----+"},{"line_number":400,"context_line":"               |                 |"},{"line_number":401,"context_line":"               |                 |"},{"line_number":402,"context_line":"               |                 |"},{"line_number":403,"context_line":"         +-----+-----------------+----+"},{"line_number":404,"context_line":"         |    port 1          port 2  |"},{"line_number":405,"context_line":"         |                            |"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_a816eeee","line":402,"range":{"start_line":402,"start_character":14,"end_line":402,"end_character":15},"updated":"2016-06-09 19:06:35.000000000","message":"We could put \"tag\u003d1\" here and \"tag\u003d2\" above port 2 to show that the local vlan mapping is applied still.","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"5538f5b25cc28c31e568eadea620fd221987b714","unresolved":false,"context_lines":[{"line_number":399,"context_line":"           +---+---+       +-----+-----+"},{"line_number":400,"context_line":"               |                 |"},{"line_number":401,"context_line":"               |                 |"},{"line_number":402,"context_line":"               |                 |"},{"line_number":403,"context_line":"         +-----+-----------------+----+"},{"line_number":404,"context_line":"         |    port 1          port 2  |"},{"line_number":405,"context_line":"         |                            |"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_bcc851a9","line":402,"range":{"start_line":402,"start_character":14,"end_line":402,"end_character":15},"in_reply_to":"7aa08908_a816eeee","updated":"2016-06-14 13:36:11.000000000","message":"thanks for the suggestion but I prefer leaving as it is","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":7448,"name":"Carl Baldwin","email":"carl@ecbaldwin.net","username":"carl-baldwin"},"change_message_id":"ad5cc759a71392ee4185b91753fe954cf5ab94da","unresolved":false,"context_lines":[{"line_number":428,"context_line":"Parent port deletion"},{"line_number":429,"context_line":"++++++++++++++++++++"},{"line_number":430,"context_line":""},{"line_number":431,"context_line":"Deleting a port that is an active parent in a trunk is forbidden. If the parent port has"},{"line_number":432,"context_line":"no trunk associated, it can be deleted. The OVS agent doesn\u0027t need to perform any action,"},{"line_number":433,"context_line":"the deletion will result in a removal of the port data from the DB."},{"line_number":434,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_8477569e","line":431,"updated":"2016-06-10 22:10:01.000000000","message":"Technically, it isn\u0027t a parent port at this point, right?  Its just a normal port.","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"5538f5b25cc28c31e568eadea620fd221987b714","unresolved":false,"context_lines":[{"line_number":428,"context_line":"Parent port deletion"},{"line_number":429,"context_line":"++++++++++++++++++++"},{"line_number":430,"context_line":""},{"line_number":431,"context_line":"Deleting a port that is an active parent in a trunk is forbidden. If the parent port has"},{"line_number":432,"context_line":"no trunk associated, it can be deleted. The OVS agent doesn\u0027t need to perform any action,"},{"line_number":433,"context_line":"the deletion will result in a removal of the port data from the DB."},{"line_number":434,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_1cb88510","line":431,"in_reply_to":"7aa08908_8477569e","updated":"2016-06-14 13:36:11.000000000","message":"yes, I will rephrase.","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"5f2a48a56469f2a042ead7b55cbfe6ff908d9db1","unresolved":false,"context_lines":[{"line_number":443,"context_line":"will detect that deletion and notify the Neutron server that the parent port is down."},{"line_number":444,"context_line":"The OVS agent will clean up the corresponding subports as explained in the next paragraph."},{"line_number":445,"context_line":""},{"line_number":446,"context_line":"When a trunk port that it\u0027s used by a VM is deleted through the API, Neutron server"},{"line_number":447,"context_line":"will send a \u0027network-vif-deleted\u0027 event to Nova that will unplug the vif from br-int"},{"line_number":448,"context_line":"and delete the firewall bridge (if appliable). The OVS agent will notice that the parent"},{"line_number":449,"context_line":"port was deleted (either by monitoring br-int or thanks the to port delete notification)"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_03a46705","line":446,"range":{"start_line":446,"start_character":7,"end_line":446,"end_character":40},"updated":"2016-06-09 10:01:48.000000000","message":"I think this is not possible in one step. Deleting the parent port of a trunk is forbidden.\n\nFirst we can delete the trunk. We leave behind:\n\n* the parent port as a regular port, still used by the VM\n\n* the child ports as regular ports, unused\n\nTherefore neutron should not send a network-vif-deleted event at this point. Nova should not notice that the trunk was deleted. \n\nI think subport wirings should be cleaned up right away after trunk deletion.\n\nAfter all this, every port (that was a parent or a child before) is just a regular port, their deletion should not be special.","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"185dbe523a2e243a01e3aa5cf06a93a9a0efa175","unresolved":false,"context_lines":[{"line_number":443,"context_line":"will detect that deletion and notify the Neutron server that the parent port is down."},{"line_number":444,"context_line":"The OVS agent will clean up the corresponding subports as explained in the next paragraph."},{"line_number":445,"context_line":""},{"line_number":446,"context_line":"When a trunk port that it\u0027s used by a VM is deleted through the API, Neutron server"},{"line_number":447,"context_line":"will send a \u0027network-vif-deleted\u0027 event to Nova that will unplug the vif from br-int"},{"line_number":448,"context_line":"and delete the firewall bridge (if appliable). The OVS agent will notice that the parent"},{"line_number":449,"context_line":"port was deleted (either by monitoring br-int or thanks the to port delete notification)"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_0d175455","line":446,"range":{"start_line":446,"start_character":7,"end_line":446,"end_character":40},"in_reply_to":"7aa08908_03a46705","updated":"2016-06-09 19:06:35.000000000","message":"When the trunk is deleted, the subports will also be deleted. The OVS agent needs to be notified of this event so that it can delete the subport vlan interfaces on the tap. The tap will then just be a regular port after that. Once it\u0027s a regular port, the standard port-deletion rules should apply (is Nova notification already the standard there?).","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"c786d2f73a8931a2a34bf5b950122e25fec95064","unresolved":false,"context_lines":[{"line_number":443,"context_line":"will detect that deletion and notify the Neutron server that the parent port is down."},{"line_number":444,"context_line":"The OVS agent will clean up the corresponding subports as explained in the next paragraph."},{"line_number":445,"context_line":""},{"line_number":446,"context_line":"When a trunk port that it\u0027s used by a VM is deleted through the API, Neutron server"},{"line_number":447,"context_line":"will send a \u0027network-vif-deleted\u0027 event to Nova that will unplug the vif from br-int"},{"line_number":448,"context_line":"and delete the firewall bridge (if appliable). The OVS agent will notice that the parent"},{"line_number":449,"context_line":"port was deleted (either by monitoring br-int or thanks the to port delete notification)"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_8d25dac3","line":446,"range":{"start_line":446,"start_character":7,"end_line":446,"end_character":40},"in_reply_to":"7aa08908_0d175455","updated":"2016-06-10 12:29:36.000000000","message":"I think we are talking about the same.","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":7448,"name":"Carl Baldwin","email":"carl@ecbaldwin.net","username":"carl-baldwin"},"change_message_id":"ad5cc759a71392ee4185b91753fe954cf5ab94da","unresolved":false,"context_lines":[{"line_number":443,"context_line":"will detect that deletion and notify the Neutron server that the parent port is down."},{"line_number":444,"context_line":"The OVS agent will clean up the corresponding subports as explained in the next paragraph."},{"line_number":445,"context_line":""},{"line_number":446,"context_line":"When a trunk port that it\u0027s used by a VM is deleted through the API, Neutron server"},{"line_number":447,"context_line":"will send a \u0027network-vif-deleted\u0027 event to Nova that will unplug the vif from br-int"},{"line_number":448,"context_line":"and delete the firewall bridge (if appliable). The OVS agent will notice that the parent"},{"line_number":449,"context_line":"port was deleted (either by monitoring br-int or thanks the to port delete notification)"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_c45b1ef5","line":446,"in_reply_to":"7aa08908_8d25dac3","updated":"2016-06-10 22:10:01.000000000","message":"I guess leaving behind parent ports when a trunk is deleted is okay with this implementation.  But, would it be okay with Option C?  In that case, trunk ports are wired up totally differently.  If you delete the trunk, then the parent port just looks like a normal port.  Yet, it will be wired up differently.\n\nDo we want this to be backend dependent?  Or, should we just prevent deleting a trunk where the port is active?","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"5538f5b25cc28c31e568eadea620fd221987b714","unresolved":false,"context_lines":[{"line_number":443,"context_line":"will detect that deletion and notify the Neutron server that the parent port is down."},{"line_number":444,"context_line":"The OVS agent will clean up the corresponding subports as explained in the next paragraph."},{"line_number":445,"context_line":""},{"line_number":446,"context_line":"When a trunk port that it\u0027s used by a VM is deleted through the API, Neutron server"},{"line_number":447,"context_line":"will send a \u0027network-vif-deleted\u0027 event to Nova that will unplug the vif from br-int"},{"line_number":448,"context_line":"and delete the firewall bridge (if appliable). The OVS agent will notice that the parent"},{"line_number":449,"context_line":"port was deleted (either by monitoring br-int or thanks the to port delete notification)"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_3c058113","line":446,"in_reply_to":"7aa08908_c45b1ef5","updated":"2016-06-14 13:36:11.000000000","message":"@Rawlin the notification network-vif-deleted is the standard. When a \"normal\" port that it\u0027s used by a VM is deleted through the API, this is exactly what happens. Neutron sends the network-vif-deleted notification to Nova that unplugs the vif from br-int and deletes the firewall bridge. \n@Carl preventing the deletion looks like the easier way. In the implementation that I proposed here there\u0027s an inconsistency: we get a request to delete a trunk but Neutron should send a network-vif-deleted event using the parent port ID, since nova doesn\u0027t know anything about the trunk. I will modify it.","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"5f2a48a56469f2a042ead7b55cbfe6ff908d9db1","unresolved":false,"context_lines":[{"line_number":445,"context_line":""},{"line_number":446,"context_line":"When a trunk port that it\u0027s used by a VM is deleted through the API, Neutron server"},{"line_number":447,"context_line":"will send a \u0027network-vif-deleted\u0027 event to Nova that will unplug the vif from br-int"},{"line_number":448,"context_line":"and delete the firewall bridge (if appliable). The OVS agent will notice that the parent"},{"line_number":449,"context_line":"port was deleted (either by monitoring br-int or thanks the to port delete notification)"},{"line_number":450,"context_line":"and it will be able to clean up the corresponing subports as explained in"},{"line_number":451,"context_line":"the next paragraph."}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_436b0fb7","line":448,"range":{"start_line":448,"start_character":35,"end_line":448,"end_character":44},"updated":"2016-06-09 10:01:48.000000000","message":"nit: applicable","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":4187,"name":"Ryan Tidwell","email":"rtidwell@suse.com","username":"ryan-tidwell"},"change_message_id":"cbb7a818b0f3765d6070da1ee7911d942507c6b7","unresolved":false,"context_lines":[{"line_number":454,"context_line":"++++++++++++++++"},{"line_number":455,"context_line":""},{"line_number":456,"context_line":"Removing a subport that is associated with a parent port that was not used to boot any"},{"line_number":457,"context_line":"VM is a no op."},{"line_number":458,"context_line":"When a subport associated with a parent port that was used to boot a VM is deleted,"},{"line_number":459,"context_line":"the OVS agent will take care of removing the VLAN interface, the firewall bridge"},{"line_number":460,"context_line":"if using iptables firewall and the port on br-int."}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_2ffda83b","line":457,"updated":"2016-06-10 17:40:19.000000000","message":"From an agent perspective, yes. There\u0027s still activity happening server-side, but perhaps that\u0027s understood and I\u0027m just being pedantic.","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"5538f5b25cc28c31e568eadea620fd221987b714","unresolved":false,"context_lines":[{"line_number":454,"context_line":"++++++++++++++++"},{"line_number":455,"context_line":""},{"line_number":456,"context_line":"Removing a subport that is associated with a parent port that was not used to boot any"},{"line_number":457,"context_line":"VM is a no op."},{"line_number":458,"context_line":"When a subport associated with a parent port that was used to boot a VM is deleted,"},{"line_number":459,"context_line":"the OVS agent will take care of removing the VLAN interface, the firewall bridge"},{"line_number":460,"context_line":"if using iptables firewall and the port on br-int."}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_7f4ffb6a","line":457,"in_reply_to":"7aa08908_2ffda83b","updated":"2016-06-14 13:36:11.000000000","message":"Done","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"5f2a48a56469f2a042ead7b55cbfe6ff908d9db1","unresolved":false,"context_lines":[{"line_number":455,"context_line":""},{"line_number":456,"context_line":"Removing a subport that is associated with a parent port that was not used to boot any"},{"line_number":457,"context_line":"VM is a no op."},{"line_number":458,"context_line":"When a subport associated with a parent port that was used to boot a VM is deleted,"},{"line_number":459,"context_line":"the OVS agent will take care of removing the VLAN interface, the firewall bridge"},{"line_number":460,"context_line":"if using iptables firewall and the port on br-int."},{"line_number":461,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_5ec2c853","line":458,"range":{"start_line":458,"start_character":45,"end_line":458,"end_character":82},"updated":"2016-06-09 10:01:48.000000000","message":"or added to a running VM any time later","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"185dbe523a2e243a01e3aa5cf06a93a9a0efa175","unresolved":false,"context_lines":[{"line_number":455,"context_line":""},{"line_number":456,"context_line":"Removing a subport that is associated with a parent port that was not used to boot any"},{"line_number":457,"context_line":"VM is a no op."},{"line_number":458,"context_line":"When a subport associated with a parent port that was used to boot a VM is deleted,"},{"line_number":459,"context_line":"the OVS agent will take care of removing the VLAN interface, the firewall bridge"},{"line_number":460,"context_line":"if using iptables firewall and the port on br-int."},{"line_number":461,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_a8b54e32","line":458,"range":{"start_line":458,"start_character":45,"end_line":458,"end_character":82},"in_reply_to":"7aa08908_5ec2c853","updated":"2016-06-09 19:06:35.000000000","message":"\u003e or added to a running VM any time later\n\nI\u0027m not sure what you mean here. This section is about subport deletion, so what does adding a subport have to do with deletion?","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"c786d2f73a8931a2a34bf5b950122e25fec95064","unresolved":false,"context_lines":[{"line_number":455,"context_line":""},{"line_number":456,"context_line":"Removing a subport that is associated with a parent port that was not used to boot any"},{"line_number":457,"context_line":"VM is a no op."},{"line_number":458,"context_line":"When a subport associated with a parent port that was used to boot a VM is deleted,"},{"line_number":459,"context_line":"the OVS agent will take care of removing the VLAN interface, the firewall bridge"},{"line_number":460,"context_line":"if using iptables firewall and the port on br-int."},{"line_number":461,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_f084355e","line":458,"range":{"start_line":458,"start_character":45,"end_line":458,"end_character":82},"in_reply_to":"7aa08908_a8b54e32","updated":"2016-06-10 12:29:36.000000000","message":"I think I just misunderstood the original sentence (which port exists at boot time, the parent or the child port?). Sorry for the confusion. Ignore my comment please.","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"5f2a48a56469f2a042ead7b55cbfe6ff908d9db1","unresolved":false,"context_lines":[{"line_number":471,"context_line":""},{"line_number":472,"context_line":"         +--------------------------------+"},{"line_number":473,"context_line":"         |             VM                 |"},{"line_number":474,"context_line":"         |   eth0                         |"},{"line_number":475,"context_line":"         +-----+--------------------------+"},{"line_number":476,"context_line":"               |"},{"line_number":477,"context_line":"               |"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_1e7a0024","line":474,"range":{"start_line":474,"start_character":33,"end_line":474,"end_character":41},"updated":"2016-06-09 10:01:48.000000000","message":"We could have eth0.100 here the same way as in the previous drawing.","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"185dbe523a2e243a01e3aa5cf06a93a9a0efa175","unresolved":false,"context_lines":[{"line_number":471,"context_line":""},{"line_number":472,"context_line":"         +--------------------------------+"},{"line_number":473,"context_line":"         |             VM                 |"},{"line_number":474,"context_line":"         |   eth0                         |"},{"line_number":475,"context_line":"         +-----+--------------------------+"},{"line_number":476,"context_line":"               |"},{"line_number":477,"context_line":"               |"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_c88ffa73","line":474,"range":{"start_line":474,"start_character":33,"end_line":474,"end_character":41},"in_reply_to":"7aa08908_1e7a0024","updated":"2016-06-09 19:06:35.000000000","message":"+1","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"5538f5b25cc28c31e568eadea620fd221987b714","unresolved":false,"context_lines":[{"line_number":471,"context_line":""},{"line_number":472,"context_line":"         +--------------------------------+"},{"line_number":473,"context_line":"         |             VM                 |"},{"line_number":474,"context_line":"         |   eth0                         |"},{"line_number":475,"context_line":"         +-----+--------------------------+"},{"line_number":476,"context_line":"               |"},{"line_number":477,"context_line":"               |"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_df3eefb6","line":474,"range":{"start_line":474,"start_character":33,"end_line":474,"end_character":41},"in_reply_to":"7aa08908_c88ffa73","updated":"2016-06-14 13:36:11.000000000","message":"Done","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"185dbe523a2e243a01e3aa5cf06a93a9a0efa175","unresolved":false,"context_lines":[{"line_number":483,"context_line":"         +-----+-----------------+--------+"},{"line_number":484,"context_line":"               |                 |"},{"line_number":485,"context_line":"               |                 |"},{"line_number":486,"context_line":"               |                 |"},{"line_number":487,"context_line":"         +-----+-----------------+---------+"},{"line_number":488,"context_line":"         | tp-patch-int     sp-patch-int   |"},{"line_number":489,"context_line":"         |                                 |"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_480e6ae6","line":486,"range":{"start_line":486,"start_character":13,"end_line":486,"end_character":15},"updated":"2016-06-09 19:06:35.000000000","message":"We could add \"tag\u003d1\" here above tp-patch-int and \"tag\u003d2\" above sp-patch-int to show that these ports will have the local vlan mapping. Below sp-patch-trunk, we could have \"tag\u003d123\" to show that the subport\u0027s tag on the trunk bridge comes from the subport\u0027s segmentation_id.","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"5538f5b25cc28c31e568eadea620fd221987b714","unresolved":false,"context_lines":[{"line_number":483,"context_line":"         +-----+-----------------+--------+"},{"line_number":484,"context_line":"               |                 |"},{"line_number":485,"context_line":"               |                 |"},{"line_number":486,"context_line":"               |                 |"},{"line_number":487,"context_line":"         +-----+-----------------+---------+"},{"line_number":488,"context_line":"         | tp-patch-int     sp-patch-int   |"},{"line_number":489,"context_line":"         |                                 |"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_3fb8d329","line":486,"range":{"start_line":486,"start_character":13,"end_line":486,"end_character":15},"in_reply_to":"7aa08908_480e6ae6","updated":"2016-06-14 13:36:11.000000000","message":"I prefer to leave the vlan tags outside of the graph.","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"5f2a48a56469f2a042ead7b55cbfe6ff908d9db1","unresolved":false,"context_lines":[{"line_number":492,"context_line":""},{"line_number":493,"context_line":"tp-patch-trunk: trunk bridge side of the patch port that implements the trunk port"},{"line_number":494,"context_line":"tp-patch-int: int bridge side of the patch port that implements the trunk port"},{"line_number":495,"context_line":"sp-patch-trunk: trunk bridge side of the patch port that implements the subport"},{"line_number":496,"context_line":"sp-patch-int: int bridge side of the patch port that implements the subport"},{"line_number":497,"context_line":""},{"line_number":498,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_7eace48b","line":495,"range":{"start_line":495,"start_character":68,"end_line":495,"end_character":71},"updated":"2016-06-09 10:01:48.000000000","message":"s/the/a/","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"5538f5b25cc28c31e568eadea620fd221987b714","unresolved":false,"context_lines":[{"line_number":492,"context_line":""},{"line_number":493,"context_line":"tp-patch-trunk: trunk bridge side of the patch port that implements the trunk port"},{"line_number":494,"context_line":"tp-patch-int: int bridge side of the patch port that implements the trunk port"},{"line_number":495,"context_line":"sp-patch-trunk: trunk bridge side of the patch port that implements the subport"},{"line_number":496,"context_line":"sp-patch-int: int bridge side of the patch port that implements the subport"},{"line_number":497,"context_line":""},{"line_number":498,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_3f8f3376","line":495,"range":{"start_line":495,"start_character":68,"end_line":495,"end_character":71},"in_reply_to":"7aa08908_7eace48b","updated":"2016-06-14 13:36:11.000000000","message":"Done","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"5f2a48a56469f2a042ead7b55cbfe6ff908d9db1","unresolved":false,"context_lines":[{"line_number":493,"context_line":"tp-patch-trunk: trunk bridge side of the patch port that implements the trunk port"},{"line_number":494,"context_line":"tp-patch-int: int bridge side of the patch port that implements the trunk port"},{"line_number":495,"context_line":"sp-patch-trunk: trunk bridge side of the patch port that implements the subport"},{"line_number":496,"context_line":"sp-patch-int: int bridge side of the patch port that implements the subport"},{"line_number":497,"context_line":""},{"line_number":498,"context_line":""},{"line_number":499,"context_line":"Trunk port creation"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_9ebb90c3","line":496,"range":{"start_line":496,"start_character":64,"end_line":496,"end_character":67},"updated":"2016-06-09 10:01:48.000000000","message":"s/the/a/","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"5538f5b25cc28c31e568eadea620fd221987b714","unresolved":false,"context_lines":[{"line_number":493,"context_line":"tp-patch-trunk: trunk bridge side of the patch port that implements the trunk port"},{"line_number":494,"context_line":"tp-patch-int: int bridge side of the patch port that implements the trunk port"},{"line_number":495,"context_line":"sp-patch-trunk: trunk bridge side of the patch port that implements the subport"},{"line_number":496,"context_line":"sp-patch-int: int bridge side of the patch port that implements the subport"},{"line_number":497,"context_line":""},{"line_number":498,"context_line":""},{"line_number":499,"context_line":"Trunk port creation"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_7feb7b13","line":496,"range":{"start_line":496,"start_character":64,"end_line":496,"end_character":67},"in_reply_to":"7aa08908_9ebb90c3","updated":"2016-06-14 13:36:11.000000000","message":"Done","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"5f2a48a56469f2a042ead7b55cbfe6ff908d9db1","unresolved":false,"context_lines":[{"line_number":500,"context_line":"+++++++++++++++++++"},{"line_number":501,"context_line":""},{"line_number":502,"context_line":"A VM is spawned passing to Nova the port-id of a parent port associated with"},{"line_number":503,"context_line":"a trunk port and the bridge to use as part of the vif details."},{"line_number":504,"context_line":"The os-vif driver creates the trunk bridge br-trunk-1 if it does not exist in plug()"},{"line_number":505,"context_line":"with the parent port ID in the external-ids field. It will create the tap interface"},{"line_number":506,"context_line":"tap1 and plug it into br-trunk-1 setting the parent port ID in the external-ids."}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_9e655046","line":503,"range":{"start_line":503,"start_character":17,"end_line":503,"end_character":61},"updated":"2016-06-09 10:01:48.000000000","message":"I think it would be cleaner to break this into two sentences:\n\n* the user passes a port-id to nova\n* neutron passes vif_details to nova\n\nAlso we could be explicit about that the bridge passed by neutron is the bridge we expect nova to plug to.","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"5f2a48a56469f2a042ead7b55cbfe6ff908d9db1","unresolved":false,"context_lines":[{"line_number":500,"context_line":"+++++++++++++++++++"},{"line_number":501,"context_line":""},{"line_number":502,"context_line":"A VM is spawned passing to Nova the port-id of a parent port associated with"},{"line_number":503,"context_line":"a trunk port and the bridge to use as part of the vif details."},{"line_number":504,"context_line":"The os-vif driver creates the trunk bridge br-trunk-1 if it does not exist in plug()"},{"line_number":505,"context_line":"with the parent port ID in the external-ids field. It will create the tap interface"},{"line_number":506,"context_line":"tap1 and plug it into br-trunk-1 setting the parent port ID in the external-ids."}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_5ebe48ae","line":503,"range":{"start_line":503,"start_character":2,"end_line":503,"end_character":12},"updated":"2016-06-09 10:01:48.000000000","message":"s/trunk port/trunk/","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"5538f5b25cc28c31e568eadea620fd221987b714","unresolved":false,"context_lines":[{"line_number":500,"context_line":"+++++++++++++++++++"},{"line_number":501,"context_line":""},{"line_number":502,"context_line":"A VM is spawned passing to Nova the port-id of a parent port associated with"},{"line_number":503,"context_line":"a trunk port and the bridge to use as part of the vif details."},{"line_number":504,"context_line":"The os-vif driver creates the trunk bridge br-trunk-1 if it does not exist in plug()"},{"line_number":505,"context_line":"with the parent port ID in the external-ids field. It will create the tap interface"},{"line_number":506,"context_line":"tap1 and plug it into br-trunk-1 setting the parent port ID in the external-ids."}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_9f42a70d","line":503,"range":{"start_line":503,"start_character":17,"end_line":503,"end_character":61},"in_reply_to":"7aa08908_9e655046","updated":"2016-06-14 13:36:11.000000000","message":"Done","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"5f2a48a56469f2a042ead7b55cbfe6ff908d9db1","unresolved":false,"context_lines":[{"line_number":503,"context_line":"a trunk port and the bridge to use as part of the vif details."},{"line_number":504,"context_line":"The os-vif driver creates the trunk bridge br-trunk-1 if it does not exist in plug()"},{"line_number":505,"context_line":"with the parent port ID in the external-ids field. It will create the tap interface"},{"line_number":506,"context_line":"tap1 and plug it into br-trunk-1 setting the parent port ID in the external-ids."},{"line_number":507,"context_line":"The OVS agent will be monitoring the creation of new trunk bridges. When it detects"},{"line_number":508,"context_line":"that a new trunk bridge has been created, it will do the following:"},{"line_number":509,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_79c35e38","line":506,"range":{"start_line":506,"start_character":67,"end_line":506,"end_character":79},"updated":"2016-06-09 10:01:48.000000000","message":"I think on the trunk tap interface we need these external_ids:\n\n {iface-id\u003d\"PARENT-PORT-ID\",\n  iface-status\u003dactive}\n\nBut we do not need \u0027attached-mac\u0027 and \u0027vm-id\u0027.","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"5538f5b25cc28c31e568eadea620fd221987b714","unresolved":false,"context_lines":[{"line_number":503,"context_line":"a trunk port and the bridge to use as part of the vif details."},{"line_number":504,"context_line":"The os-vif driver creates the trunk bridge br-trunk-1 if it does not exist in plug()"},{"line_number":505,"context_line":"with the parent port ID in the external-ids field. It will create the tap interface"},{"line_number":506,"context_line":"tap1 and plug it into br-trunk-1 setting the parent port ID in the external-ids."},{"line_number":507,"context_line":"The OVS agent will be monitoring the creation of new trunk bridges. When it detects"},{"line_number":508,"context_line":"that a new trunk bridge has been created, it will do the following:"},{"line_number":509,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_ea5b2752","line":506,"range":{"start_line":506,"start_character":67,"end_line":506,"end_character":79},"in_reply_to":"7aa08908_79c35e38","updated":"2016-06-14 13:36:11.000000000","message":"This is set in nova, I would avoid changing anything there if not really needed.","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"5f2a48a56469f2a042ead7b55cbfe6ff908d9db1","unresolved":false,"context_lines":[{"line_number":504,"context_line":"The os-vif driver creates the trunk bridge br-trunk-1 if it does not exist in plug()"},{"line_number":505,"context_line":"with the parent port ID in the external-ids field. It will create the tap interface"},{"line_number":506,"context_line":"tap1 and plug it into br-trunk-1 setting the parent port ID in the external-ids."},{"line_number":507,"context_line":"The OVS agent will be monitoring the creation of new trunk bridges. When it detects"},{"line_number":508,"context_line":"that a new trunk bridge has been created, it will do the following:"},{"line_number":509,"context_line":""},{"line_number":510,"context_line":"ovs-vsctl add-port  br-trunk-1 tp-patch-trunk -- set Interface tp-patch-trunk type\u003dpatch options:peer\u003dtp-patch-int"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_f4075315","line":507,"range":{"start_line":507,"start_character":0,"end_line":507,"end_character":68},"updated":"2016-06-09 10:01:48.000000000","message":"I was thinking about this while looking at the ovs-agent code. Today polling_manager is all about monitoring the Interface table in ovs db. The actual command is this:\n\n ovsdb-client monitor Interface name,ofport,external_ids --format\u003djson\n\nIf we want to monitor the Bridge table too, then we can\n\n* either start two ovsdb-clients, one for each table, but then we lose the ordering of events between bridges and interfaces\n* or start one ovsdb-client (monitor ALL), but then start filtering a lot of noise coming from ovs db.\n\nSo what if instead of monitoring the bridge we would monitor the trunk tap in that bridge? Seems simpler to implement while losing nothing. When we need to know if a new interface is a trunk or a regular interface, we could:\n\n* either look up which bridge it is in,\n* or better add an extra iface-type\u003dtrunk to its external_ids.\n\nWhat do you think?","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"8692b0e4f8d42ca40e09571f785f1826f209ece4","unresolved":false,"context_lines":[{"line_number":504,"context_line":"The os-vif driver creates the trunk bridge br-trunk-1 if it does not exist in plug()"},{"line_number":505,"context_line":"with the parent port ID in the external-ids field. It will create the tap interface"},{"line_number":506,"context_line":"tap1 and plug it into br-trunk-1 setting the parent port ID in the external-ids."},{"line_number":507,"context_line":"The OVS agent will be monitoring the creation of new trunk bridges. When it detects"},{"line_number":508,"context_line":"that a new trunk bridge has been created, it will do the following:"},{"line_number":509,"context_line":""},{"line_number":510,"context_line":"ovs-vsctl add-port  br-trunk-1 tp-patch-trunk -- set Interface tp-patch-trunk type\u003dpatch options:peer\u003dtp-patch-int"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_97ea67d6","line":507,"range":{"start_line":507,"start_character":0,"end_line":507,"end_character":68},"in_reply_to":"7aa08908_07114152","updated":"2016-06-14 20:10:02.000000000","message":"\u003e \u003e I believe the polling manager just tells the agent whether or not new interfaces have been added. After that, it\u0027s up to the agent to delve deeper into br-int and the ancillary bridges (i.e. br-ex) to see if new ports were added to those bridges.\n\n\u003e That\u0027s not correct, the polling manager passes an event list to the OVS agent specifying which ports have been added/removed.\n\nYes, I spoke too soon, because either the polling manager has a get_events() method or not. This comes down to whether or not \"minimize_polling\" is set to True or False, respectively. If True, then it will provide a list of ports added/removed. If False, then it scans the ports on br-int and compares them to a list of current ports that the agent maintains. This is True by default but could be set to False, and I don\u0027t see how we can monitor for new trunk ports without the polling manager\u0027s events.\n\nHow will we handle that case?","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"c786d2f73a8931a2a34bf5b950122e25fec95064","unresolved":false,"context_lines":[{"line_number":504,"context_line":"The os-vif driver creates the trunk bridge br-trunk-1 if it does not exist in plug()"},{"line_number":505,"context_line":"with the parent port ID in the external-ids field. It will create the tap interface"},{"line_number":506,"context_line":"tap1 and plug it into br-trunk-1 setting the parent port ID in the external-ids."},{"line_number":507,"context_line":"The OVS agent will be monitoring the creation of new trunk bridges. When it detects"},{"line_number":508,"context_line":"that a new trunk bridge has been created, it will do the following:"},{"line_number":509,"context_line":""},{"line_number":510,"context_line":"ovs-vsctl add-port  br-trunk-1 tp-patch-trunk -- set Interface tp-patch-trunk type\u003dpatch options:peer\u003dtp-patch-int"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_30019d6b","line":507,"range":{"start_line":507,"start_character":0,"end_line":507,"end_character":68},"in_reply_to":"7aa08908_1558fa55","updated":"2016-06-10 12:29:36.000000000","message":"It\u0027s a very good point to think about where we break off from the current agent code and write a new code path to handle trunk ports. I have never felt confident modifying ovs-agent logic without accidentally breaking it. I should get acquainted with l2 agent extensions. Please share the code if you can.","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":4187,"name":"Ryan Tidwell","email":"rtidwell@suse.com","username":"ryan-tidwell"},"change_message_id":"cbb7a818b0f3765d6070da1ee7911d942507c6b7","unresolved":false,"context_lines":[{"line_number":504,"context_line":"The os-vif driver creates the trunk bridge br-trunk-1 if it does not exist in plug()"},{"line_number":505,"context_line":"with the parent port ID in the external-ids field. It will create the tap interface"},{"line_number":506,"context_line":"tap1 and plug it into br-trunk-1 setting the parent port ID in the external-ids."},{"line_number":507,"context_line":"The OVS agent will be monitoring the creation of new trunk bridges. When it detects"},{"line_number":508,"context_line":"that a new trunk bridge has been created, it will do the following:"},{"line_number":509,"context_line":""},{"line_number":510,"context_line":"ovs-vsctl add-port  br-trunk-1 tp-patch-trunk -- set Interface tp-patch-trunk type\u003dpatch options:peer\u003dtp-patch-int"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_ea5a6ef1","line":507,"range":{"start_line":507,"start_character":0,"end_line":507,"end_character":68},"in_reply_to":"7aa08908_1558fa55","updated":"2016-06-10 17:40:19.000000000","message":"The more I think about it, the more I\u0027m drawn to the \"Nova plugs first patch port pair\" approach.  It does mean we have to pass a little more information to Nova such as use_veth_interconnection. However, as Rawlin points out it does not require us to poll for different bridges and ports. It seems like the existing polling loop would pick up a new port on br-int, and from there we could delegate to an l2 agent extension to handle the rest of the trunk/subport configuration. This has the nice property of allowing us to handle trunk-specific logic with code that is nicely de-coupled from the main agent.\n\nOf course the devil is now in the details on the os-vif side, but it seems like passing just 1 or 2 fields in trunk_details or vif_details should provide os-vif with all the data it needs to work with.","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":4187,"name":"Ryan Tidwell","email":"rtidwell@suse.com","username":"ryan-tidwell"},"change_message_id":"cbb7a818b0f3765d6070da1ee7911d942507c6b7","unresolved":false,"context_lines":[{"line_number":504,"context_line":"The os-vif driver creates the trunk bridge br-trunk-1 if it does not exist in plug()"},{"line_number":505,"context_line":"with the parent port ID in the external-ids field. It will create the tap interface"},{"line_number":506,"context_line":"tap1 and plug it into br-trunk-1 setting the parent port ID in the external-ids."},{"line_number":507,"context_line":"The OVS agent will be monitoring the creation of new trunk bridges. When it detects"},{"line_number":508,"context_line":"that a new trunk bridge has been created, it will do the following:"},{"line_number":509,"context_line":""},{"line_number":510,"context_line":"ovs-vsctl add-port  br-trunk-1 tp-patch-trunk -- set Interface tp-patch-trunk type\u003dpatch options:peer\u003dtp-patch-int"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_efad609c","line":507,"range":{"start_line":507,"start_character":0,"end_line":507,"end_character":68},"in_reply_to":"7aa08908_30019d6b","updated":"2016-06-10 17:40:19.000000000","message":"See https://github.com/openstack/neutron/blob/master/neutron/agent/l2/extensions/qos.py as an example. QoS actually seems to illustrate some nice patterns for us to consider following.","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"c133e2bca1c2a69cbca70433032762d154e92d12","unresolved":false,"context_lines":[{"line_number":504,"context_line":"The os-vif driver creates the trunk bridge br-trunk-1 if it does not exist in plug()"},{"line_number":505,"context_line":"with the parent port ID in the external-ids field. It will create the tap interface"},{"line_number":506,"context_line":"tap1 and plug it into br-trunk-1 setting the parent port ID in the external-ids."},{"line_number":507,"context_line":"The OVS agent will be monitoring the creation of new trunk bridges. When it detects"},{"line_number":508,"context_line":"that a new trunk bridge has been created, it will do the following:"},{"line_number":509,"context_line":""},{"line_number":510,"context_line":"ovs-vsctl add-port  br-trunk-1 tp-patch-trunk -- set Interface tp-patch-trunk type\u003dpatch options:peer\u003dtp-patch-int"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_ea41c03a","line":507,"range":{"start_line":507,"start_character":0,"end_line":507,"end_character":68},"in_reply_to":"7aa08908_30019d6b","updated":"2016-06-10 23:16:34.000000000","message":"https://review.openstack.org/#/c/328526/ - there is my basic trunk L2 agent extension so far. The extension itself is fairly small, yet the functionality of it could be 90% there if we currently had the RPC interface finished (the \"get subports by trunk port id\" RPC call is stubbed out). Then what\u0027s left is handling dynamic addition/removal of subports via RPC calls from the server.\n\nWe only get to use this if Nova plugs the 1st patch port pair. Hopefully now everyone can see how simple this agent implementation can be vs implementing monitoring for new trunk bridges.","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"5538f5b25cc28c31e568eadea620fd221987b714","unresolved":false,"context_lines":[{"line_number":504,"context_line":"The os-vif driver creates the trunk bridge br-trunk-1 if it does not exist in plug()"},{"line_number":505,"context_line":"with the parent port ID in the external-ids field. It will create the tap interface"},{"line_number":506,"context_line":"tap1 and plug it into br-trunk-1 setting the parent port ID in the external-ids."},{"line_number":507,"context_line":"The OVS agent will be monitoring the creation of new trunk bridges. When it detects"},{"line_number":508,"context_line":"that a new trunk bridge has been created, it will do the following:"},{"line_number":509,"context_line":""},{"line_number":510,"context_line":"ovs-vsctl add-port  br-trunk-1 tp-patch-trunk -- set Interface tp-patch-trunk type\u003dpatch options:peer\u003dtp-patch-int"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_07114152","line":507,"range":{"start_line":507,"start_character":0,"end_line":507,"end_character":68},"in_reply_to":"7aa08908_51a78ba6","updated":"2016-06-14 13:36:11.000000000","message":"@Rawlin \n\u003e I believe the polling manager just tells the agent whether or not new interfaces have been added. After that, it\u0027s up to the agent to delve deeper into br-int and the ancillary bridges (i.e. br-ex) to see if new ports were added to those bridges.\n\nThat\u0027s not correct, the polling manager passes an event list to the OVS agent specifying which ports have been added/removed.\n\nLooking at the discussion in [1] and to the comments here I think this part can stay as it is. I will provide a WIP implementation shortly to give an idea of how it would look like.\n\n[1] http://lists.openstack.org/pipermail/openstack-dev/2016-June/097025.html","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":4187,"name":"Ryan Tidwell","email":"rtidwell@suse.com","username":"ryan-tidwell"},"change_message_id":"ed25dffc421f4ed87dbf0ae0e45fd26baaed6d59","unresolved":false,"context_lines":[{"line_number":504,"context_line":"The os-vif driver creates the trunk bridge br-trunk-1 if it does not exist in plug()"},{"line_number":505,"context_line":"with the parent port ID in the external-ids field. It will create the tap interface"},{"line_number":506,"context_line":"tap1 and plug it into br-trunk-1 setting the parent port ID in the external-ids."},{"line_number":507,"context_line":"The OVS agent will be monitoring the creation of new trunk bridges. When it detects"},{"line_number":508,"context_line":"that a new trunk bridge has been created, it will do the following:"},{"line_number":509,"context_line":""},{"line_number":510,"context_line":"ovs-vsctl add-port  br-trunk-1 tp-patch-trunk -- set Interface tp-patch-trunk type\u003dpatch options:peer\u003dtp-patch-int"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_a76d3106","line":507,"in_reply_to":"7aa08908_b183b3cb","updated":"2016-06-10 23:15:14.000000000","message":"I hope nobody thinks I\u0027m blowing off passing information to Nova as trivial. It\u0027s not, and we want to think long and hard about this interaction. There are a couple things that stick out to me:\n\n1. It\u0027s been pointed out in previous comments, if we embrace the philosophy that Nova deals with L1 and Neutron handles L2 up in the stack passing use_veth_connection is an important detail for os_vif to have.\n\n2. When compared to the effort of adding specialized OVS polling mechanisms just for trunk-related items, I find a solution that relies on os-vif to handle the initial plugging into br-int the least invasive to the OVS agent. I also see it mirroring very closely the workflow we already have in place for monitoring newly plugged vif\u0027s.\n\nI\u0027m just trying to be practical and I\u0027m speaking in relative terms about compolexity. There\u0027s nothing simple about the Nova-Neutron interaction. But all things considered, I would rather Neutron send extra information over the wall and let os-vif run with it in this case. It won\u0027t be trivial, but compared with the alternative I feel better about this approach.","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"1038073a40a5b0d96f1382af5ad195c7e740f228","unresolved":false,"context_lines":[{"line_number":504,"context_line":"The os-vif driver creates the trunk bridge br-trunk-1 if it does not exist in plug()"},{"line_number":505,"context_line":"with the parent port ID in the external-ids field. It will create the tap interface"},{"line_number":506,"context_line":"tap1 and plug it into br-trunk-1 setting the parent port ID in the external-ids."},{"line_number":507,"context_line":"The OVS agent will be monitoring the creation of new trunk bridges. When it detects"},{"line_number":508,"context_line":"that a new trunk bridge has been created, it will do the following:"},{"line_number":509,"context_line":""},{"line_number":510,"context_line":"ovs-vsctl add-port  br-trunk-1 tp-patch-trunk -- set Interface tp-patch-trunk type\u003dpatch options:peer\u003dtp-patch-int"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_51a78ba6","line":507,"range":{"start_line":507,"start_character":0,"end_line":507,"end_character":68},"in_reply_to":"7aa08908_ea41c03a","updated":"2016-06-14 10:54:58.000000000","message":"For some time, I have been leaning on the option of encapsulating the least amount of implementation details into os-vif, and thus I\u0027d rather keep os-vif logic limited to just creating the bridge if missing, and plugging the VM\u0027s interface.\n\nI think the argument of choosing one approach over the other can be turned into a religious debate, but ultimately the actual data plane wiring doesn\u0027t change, and on that basis, I think the approach where Neutron is more widely in control should be preferred.","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":7448,"name":"Carl Baldwin","email":"carl@ecbaldwin.net","username":"carl-baldwin"},"change_message_id":"ad5cc759a71392ee4185b91753fe954cf5ab94da","unresolved":false,"context_lines":[{"line_number":504,"context_line":"The os-vif driver creates the trunk bridge br-trunk-1 if it does not exist in plug()"},{"line_number":505,"context_line":"with the parent port ID in the external-ids field. It will create the tap interface"},{"line_number":506,"context_line":"tap1 and plug it into br-trunk-1 setting the parent port ID in the external-ids."},{"line_number":507,"context_line":"The OVS agent will be monitoring the creation of new trunk bridges. When it detects"},{"line_number":508,"context_line":"that a new trunk bridge has been created, it will do the following:"},{"line_number":509,"context_line":""},{"line_number":510,"context_line":"ovs-vsctl add-port  br-trunk-1 tp-patch-trunk -- set Interface tp-patch-trunk type\u003dpatch options:peer\u003dtp-patch-int"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_b183b3cb","line":507,"in_reply_to":"7aa08908_ea5a6ef1","updated":"2016-06-10 22:10:01.000000000","message":"I wouldn\u0027t underestimate sending something like use_veth_interconnection over to Nova.  In general, we shouldn\u0027t leak details that we don\u0027t need to.  I understand that we have to work with what we have and there are cons to the other approach.  I\u0027m just saying don\u0027t blow it off as something trivial.\n\nI tend to think we already leak a little too much over in to Nova.  I think Neutron should own any code that creates bridges, patch ports, veths and anything like that and provide libraries that abstract it (and could read neutron config if necessary).  But, that ship has sailed.  BTW, I don\u0027t think os-vif repo is quite what I\u0027m looking for here since it seems to still combine Nova and Neutron in to a repo that isn\u0027t really part of either.","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"185dbe523a2e243a01e3aa5cf06a93a9a0efa175","unresolved":false,"context_lines":[{"line_number":504,"context_line":"The os-vif driver creates the trunk bridge br-trunk-1 if it does not exist in plug()"},{"line_number":505,"context_line":"with the parent port ID in the external-ids field. It will create the tap interface"},{"line_number":506,"context_line":"tap1 and plug it into br-trunk-1 setting the parent port ID in the external-ids."},{"line_number":507,"context_line":"The OVS agent will be monitoring the creation of new trunk bridges. When it detects"},{"line_number":508,"context_line":"that a new trunk bridge has been created, it will do the following:"},{"line_number":509,"context_line":""},{"line_number":510,"context_line":"ovs-vsctl add-port  br-trunk-1 tp-patch-trunk -- set Interface tp-patch-trunk type\u003dpatch options:peer\u003dtp-patch-int"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_1558fa55","line":507,"range":{"start_line":507,"start_character":0,"end_line":507,"end_character":68},"in_reply_to":"7aa08908_f4075315","updated":"2016-06-09 19:06:35.000000000","message":"Apologies for the long comment, but the pros and cons of the vif-plugging implementation need to be seriously discussed because it will have major effects on the agent implementation.\n\nI believe there is more to the OVS agent polling for new interfaces than just the polling_manager. I believe the polling manager just tells the agent whether or not new interfaces have been added. After that, it\u0027s up to the agent to delve deeper into br-int and the ancillary bridges (i.e. br-ex) to see if new ports were added to those bridges. These OVSBridge objects are initialized upon agent startup and are pretty hardcoded in the agent as \"the bridges to check for new ports on\".\n\nI believe it will be a hefty addition to the agent to implement the current \"port monitoring\" on trunk bridges that are dynamically created/deleted. That will also require lots of direct modifications to the current code path rather than assuming an entirely new code path to handle trunks.\n\nIn order to reduce the chances of introducing a regression in the agent, this functionality should be implemented either as a separate code path or an L2 agent extension to be safe.\n\nIf we go the \"monitor for new bridges\" route, that would ideally be implemented as a completely new code path in the agent or as an L2 agent extension (the L2 agent extension framework would have to be modified to handle this, which might also require another neutron-spec).\n\nThis brings me to \u0027pros\u0027 of the \"have Nova create the 1st patch port pair\" approach:\n1) The trunk implementation in the agent can be simply and cleanly implemented as an L2 agent extension (with the current framework as-is). I\u0027ve started a basic L2 agent extension implementation which I\u0027d be happy to share just to show you how much simpler this is. By having Nova add a port to br-int like it currently does with everything else, we get SO MUCH for free.\n2) It doesn\u0027t require adding a new code path directly in the agent to handle monitoring of new trunk bridges.\n3) It doesn\u0027t require modifying the current code path to support \"port monitoring\" for the dynamic creation/deletion of trunk bridges.\n4) The approach mirrors the current hybrid_ovs vif-plugging implementation in Nova.\n\nI know that one of the mentioned \u0027cons\u0027 of this approach was that Nova couldn\u0027t create veth pairs over patch port pairs if Neutron\u0027s use_veth_interconnection was set to True. However, the mech_driver could easily be modified to pass this information to Nova through the vif_details, so that Nova could optionally use a veth pair for the 1st subport connection to br-int.\n\nSo, why would we choose the approach that has a larger chance to introduce regressions, doesn\u0027t take advantage of available frameworks, introduces new code paths and more complex agent modifications that will need to be maintained, and adds further inefficiencies? Just to reduce a few lines of Nova changes we need? Just my 2 cents...","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"185dbe523a2e243a01e3aa5cf06a93a9a0efa175","unresolved":false,"context_lines":[{"line_number":508,"context_line":"that a new trunk bridge has been created, it will do the following:"},{"line_number":509,"context_line":""},{"line_number":510,"context_line":"ovs-vsctl add-port  br-trunk-1 tp-patch-trunk -- set Interface tp-patch-trunk type\u003dpatch options:peer\u003dtp-patch-int"},{"line_number":511,"context_line":"ovs-vsctl add-port br-int tp-patch-int tag\u003d3 -- set Interface tp-patch-int type\u003dpatch options:peer\u003dtp-patch-trunk"},{"line_number":512,"context_line":""},{"line_number":513,"context_line":""},{"line_number":514,"context_line":"A patch port is created to connect the trunk bridge to the integration bridge."}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_38e1add1","line":511,"range":{"start_line":511,"start_character":39,"end_line":511,"end_character":44},"updated":"2016-06-09 19:06:35.000000000","message":"This part could be handled as part of the agent\u0027s typical \"port monitoring\" process of br-int (i.e. applying the local vlan mapping, adding flows, etc.). Same goes for adding the tag on L534. I think using the existing mechanism would be best.","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"5538f5b25cc28c31e568eadea620fd221987b714","unresolved":false,"context_lines":[{"line_number":508,"context_line":"that a new trunk bridge has been created, it will do the following:"},{"line_number":509,"context_line":""},{"line_number":510,"context_line":"ovs-vsctl add-port  br-trunk-1 tp-patch-trunk -- set Interface tp-patch-trunk type\u003dpatch options:peer\u003dtp-patch-int"},{"line_number":511,"context_line":"ovs-vsctl add-port br-int tp-patch-int tag\u003d3 -- set Interface tp-patch-int type\u003dpatch options:peer\u003dtp-patch-trunk"},{"line_number":512,"context_line":""},{"line_number":513,"context_line":""},{"line_number":514,"context_line":"A patch port is created to connect the trunk bridge to the integration bridge."}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_47b3494d","line":511,"range":{"start_line":511,"start_character":39,"end_line":511,"end_character":44},"in_reply_to":"7aa08908_38e1add1","updated":"2016-06-14 13:36:11.000000000","message":"I was not assuming a new mechanism, shall I state it more clearly?","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"8692b0e4f8d42ca40e09571f785f1826f209ece4","unresolved":false,"context_lines":[{"line_number":508,"context_line":"that a new trunk bridge has been created, it will do the following:"},{"line_number":509,"context_line":""},{"line_number":510,"context_line":"ovs-vsctl add-port  br-trunk-1 tp-patch-trunk -- set Interface tp-patch-trunk type\u003dpatch options:peer\u003dtp-patch-int"},{"line_number":511,"context_line":"ovs-vsctl add-port br-int tp-patch-int tag\u003d3 -- set Interface tp-patch-int type\u003dpatch options:peer\u003dtp-patch-trunk"},{"line_number":512,"context_line":""},{"line_number":513,"context_line":""},{"line_number":514,"context_line":"A patch port is created to connect the trunk bridge to the integration bridge."}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_44396bf1","line":511,"range":{"start_line":511,"start_character":39,"end_line":511,"end_character":44},"in_reply_to":"7aa08908_47b3494d","updated":"2016-06-14 20:10:02.000000000","message":"I think so, because the local vlan tags on br-int ports aren\u0027t added in the same step that the OVS agent creates these patch port pairs. This line makes it seem like it\u0027s the same step.","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"5f2a48a56469f2a042ead7b55cbfe6ff908d9db1","unresolved":false,"context_lines":[{"line_number":516,"context_line":"tag. It will carry untagged traffic."},{"line_number":517,"context_line":"tp-patch-int, the br-int side the patch port is tagged with VLAN 3. We assume that the"},{"line_number":518,"context_line":"trunk port is on network1 that on this host is associated with VLAN 3."},{"line_number":519,"context_line":"The OVS agent will set the trunk port ID in the external-ids of tp-patch-trunk and tp-patch-int."},{"line_number":520,"context_line":"If the parent port is associated with one or more subports the agent will process them as"},{"line_number":521,"context_line":"described in the next paragraph."},{"line_number":522,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_54be27ae","line":519,"range":{"start_line":519,"start_character":0,"end_line":519,"end_character":96},"updated":"2016-06-09 10:01:48.000000000","message":"On tp-patch-int I think we should have the full set of external_ids for compatibility (even if that is somewhat redundant), eg:\n\n {attached-mac\u003d\"fa:16:3e:e7:26:11\",\n  iface-id\u003d\"04d612a6-974a-4bd3-b172-d539d13c74f7\",\n  iface-status\u003dactive,\n  vm-id\u003d\"f8b2f9aa-1aae-4b75-bd64-7dc7e67e6bc4\"}","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"8692b0e4f8d42ca40e09571f785f1826f209ece4","unresolved":false,"context_lines":[{"line_number":516,"context_line":"tag. It will carry untagged traffic."},{"line_number":517,"context_line":"tp-patch-int, the br-int side the patch port is tagged with VLAN 3. We assume that the"},{"line_number":518,"context_line":"trunk port is on network1 that on this host is associated with VLAN 3."},{"line_number":519,"context_line":"The OVS agent will set the trunk port ID in the external-ids of tp-patch-trunk and tp-patch-int."},{"line_number":520,"context_line":"If the parent port is associated with one or more subports the agent will process them as"},{"line_number":521,"context_line":"described in the next paragraph."},{"line_number":522,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_97b26747","line":519,"range":{"start_line":519,"start_character":0,"end_line":519,"end_character":96},"in_reply_to":"7aa08908_47eb49dc","updated":"2016-06-14 20:10:02.000000000","message":"I was referencing this snippet in the OVS agent:\n\n         def _process_port(port, ports, ancillary_ports):\n            # check \u0027iface-id\u0027 is set otherwise is not a port\n            # the agent should care about\n            if \u0027attached-mac\u0027 in port.get(\u0027external_ids\u0027, []):\n                iface_id \u003d self.int_br.portid_from_external_ids(\n                    port[\u0027external_ids\u0027])\n\nSo we will need \u0027attached-mac\u0027 and \u0027iface-id\u0027 (which is the port_id).","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"185dbe523a2e243a01e3aa5cf06a93a9a0efa175","unresolved":false,"context_lines":[{"line_number":516,"context_line":"tag. It will carry untagged traffic."},{"line_number":517,"context_line":"tp-patch-int, the br-int side the patch port is tagged with VLAN 3. We assume that the"},{"line_number":518,"context_line":"trunk port is on network1 that on this host is associated with VLAN 3."},{"line_number":519,"context_line":"The OVS agent will set the trunk port ID in the external-ids of tp-patch-trunk and tp-patch-int."},{"line_number":520,"context_line":"If the parent port is associated with one or more subports the agent will process them as"},{"line_number":521,"context_line":"described in the next paragraph."},{"line_number":522,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_785b352f","line":519,"range":{"start_line":519,"start_character":0,"end_line":519,"end_character":96},"in_reply_to":"7aa08908_54be27ae","updated":"2016-06-09 19:06:35.000000000","message":"I believe the OVS agent only cares about \u0027attached-mac\u0027 and \u0027iface-id\u0027 from what I can tell, but it shouldn\u0027t hurt to have it all.\n\nWe mainly need this metadata on the br-int ports so that the agent can properly configure them using the existing \"port monitoring\" mechanism (i.e. apply the local vlan mapping, flows, etc).","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"5538f5b25cc28c31e568eadea620fd221987b714","unresolved":false,"context_lines":[{"line_number":516,"context_line":"tag. It will carry untagged traffic."},{"line_number":517,"context_line":"tp-patch-int, the br-int side the patch port is tagged with VLAN 3. We assume that the"},{"line_number":518,"context_line":"trunk port is on network1 that on this host is associated with VLAN 3."},{"line_number":519,"context_line":"The OVS agent will set the trunk port ID in the external-ids of tp-patch-trunk and tp-patch-int."},{"line_number":520,"context_line":"If the parent port is associated with one or more subports the agent will process them as"},{"line_number":521,"context_line":"described in the next paragraph."},{"line_number":522,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_47eb49dc","line":519,"range":{"start_line":519,"start_character":0,"end_line":519,"end_character":96},"in_reply_to":"7aa08908_785b352f","updated":"2016-06-14 13:36:11.000000000","message":"(YAGNI) If we realize that we need one of those values we can just add it. Probably mac makes sense.","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"185dbe523a2e243a01e3aa5cf06a93a9a0efa175","unresolved":false,"context_lines":[{"line_number":531,"context_line":""},{"line_number":532,"context_line":""},{"line_number":533,"context_line":"ovs-vsctl add-port  br-trunk-1 sp-patch-trunk tag\u003d100 -- set Interface sp-patch-trunk type\u003dpatch options:peer\u003dsp-patch-int"},{"line_number":534,"context_line":"ovs-vsctl add-port br-int sp-patch-int tag\u003d5 -- set Interface sp-patch-int type\u003dpatch options:peer\u003dsp-patch-trunk"},{"line_number":535,"context_line":""},{"line_number":536,"context_line":"This patch port connects the trunk bridge to the integration bridge."},{"line_number":537,"context_line":"sp-patch-trunk, the trunk bridge side of the patch is tagged using VLAN 100."}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_38088d91","line":534,"range":{"start_line":534,"start_character":39,"end_line":534,"end_character":44},"updated":"2016-06-09 19:06:35.000000000","message":"see L511 comment.","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"5f2a48a56469f2a042ead7b55cbfe6ff908d9db1","unresolved":false,"context_lines":[{"line_number":538,"context_line":"We assume that the segmentation ID of the subport is 100."},{"line_number":539,"context_line":"sp-patch-int, the br-int side of the patch port is tagged with VLAN 5. We"},{"line_number":540,"context_line":"assume that the subport is on network2 that on this host uses VLAN 5."},{"line_number":541,"context_line":"The OVS agent will set the subport ID in the external-ids of sp-patch-trunk and sp-patch-int."},{"line_number":542,"context_line":""},{"line_number":543,"context_line":"*Inbound traffic from the VM point of view*"},{"line_number":544,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_f46bd3ee","line":541,"range":{"start_line":541,"start_character":0,"end_line":541,"end_character":93},"updated":"2016-06-09 10:01:48.000000000","message":"On sp-patch-int we should have the full set of external_ids (as in lin 519), so other parts of the agent (for example the ovs firewall driver) can\u0027t tell the difference between br-int ports belonging to regular, parent or child ports.","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"5538f5b25cc28c31e568eadea620fd221987b714","unresolved":false,"context_lines":[{"line_number":538,"context_line":"We assume that the segmentation ID of the subport is 100."},{"line_number":539,"context_line":"sp-patch-int, the br-int side of the patch port is tagged with VLAN 5. We"},{"line_number":540,"context_line":"assume that the subport is on network2 that on this host uses VLAN 5."},{"line_number":541,"context_line":"The OVS agent will set the subport ID in the external-ids of sp-patch-trunk and sp-patch-int."},{"line_number":542,"context_line":""},{"line_number":543,"context_line":"*Inbound traffic from the VM point of view*"},{"line_number":544,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_a2013b07","line":541,"range":{"start_line":541,"start_character":0,"end_line":541,"end_character":93},"in_reply_to":"7aa08908_8f765c3c","updated":"2016-06-14 13:36:11.000000000","message":"The ovs firewall driver doesn\u0027t query ovsdb but gets those info from a dictionary that it\u0027s passed by the ovs agent","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"8692b0e4f8d42ca40e09571f785f1826f209ece4","unresolved":false,"context_lines":[{"line_number":538,"context_line":"We assume that the segmentation ID of the subport is 100."},{"line_number":539,"context_line":"sp-patch-int, the br-int side of the patch port is tagged with VLAN 5. We"},{"line_number":540,"context_line":"assume that the subport is on network2 that on this host uses VLAN 5."},{"line_number":541,"context_line":"The OVS agent will set the subport ID in the external-ids of sp-patch-trunk and sp-patch-int."},{"line_number":542,"context_line":""},{"line_number":543,"context_line":"*Inbound traffic from the VM point of view*"},{"line_number":544,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_576f5f91","line":541,"range":{"start_line":541,"start_character":0,"end_line":541,"end_character":93},"in_reply_to":"7aa08908_a2013b07","updated":"2016-06-14 20:10:02.000000000","message":"We will need the external_ids on sp-patch-int so that the agent will correctly configure them (see snippet in comment on L519).","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":4187,"name":"Ryan Tidwell","email":"rtidwell@suse.com","username":"ryan-tidwell"},"change_message_id":"cbb7a818b0f3765d6070da1ee7911d942507c6b7","unresolved":false,"context_lines":[{"line_number":538,"context_line":"We assume that the segmentation ID of the subport is 100."},{"line_number":539,"context_line":"sp-patch-int, the br-int side of the patch port is tagged with VLAN 5. We"},{"line_number":540,"context_line":"assume that the subport is on network2 that on this host uses VLAN 5."},{"line_number":541,"context_line":"The OVS agent will set the subport ID in the external-ids of sp-patch-trunk and sp-patch-int."},{"line_number":542,"context_line":""},{"line_number":543,"context_line":"*Inbound traffic from the VM point of view*"},{"line_number":544,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_8f765c3c","line":541,"range":{"start_line":541,"start_character":0,"end_line":541,"end_character":93},"in_reply_to":"7aa08908_f46bd3ee","updated":"2016-06-10 17:40:19.000000000","message":"+1","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"185dbe523a2e243a01e3aa5cf06a93a9a0efa175","unresolved":false,"context_lines":[{"line_number":538,"context_line":"We assume that the segmentation ID of the subport is 100."},{"line_number":539,"context_line":"sp-patch-int, the br-int side of the patch port is tagged with VLAN 5. We"},{"line_number":540,"context_line":"assume that the subport is on network2 that on this host uses VLAN 5."},{"line_number":541,"context_line":"The OVS agent will set the subport ID in the external-ids of sp-patch-trunk and sp-patch-int."},{"line_number":542,"context_line":""},{"line_number":543,"context_line":"*Inbound traffic from the VM point of view*"},{"line_number":544,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_38936d51","line":541,"range":{"start_line":541,"start_character":0,"end_line":541,"end_character":93},"in_reply_to":"7aa08908_f46bd3ee","updated":"2016-06-09 19:06:35.000000000","message":"agreed.","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"5f2a48a56469f2a042ead7b55cbfe6ff908d9db1","unresolved":false,"context_lines":[{"line_number":560,"context_line":"Parent port deletion"},{"line_number":561,"context_line":"++++++++++++++++++++"},{"line_number":562,"context_line":""},{"line_number":563,"context_line":"Deleting a port that is an active parent in a trunk is forbidden. If the parent port has"},{"line_number":564,"context_line":"no trunk associated, it can be deleted. The OVS agent doesn\u0027t need to perform any action."},{"line_number":565,"context_line":""},{"line_number":566,"context_line":"Trunk port deletion"},{"line_number":567,"context_line":"+++++++++++++++++++"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_7457e325","line":564,"range":{"start_line":563,"start_character":66,"end_line":564,"end_character":39},"updated":"2016-06-09 10:01:48.000000000","message":"There\u0027s no such thing.","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"185dbe523a2e243a01e3aa5cf06a93a9a0efa175","unresolved":false,"context_lines":[{"line_number":560,"context_line":"Parent port deletion"},{"line_number":561,"context_line":"++++++++++++++++++++"},{"line_number":562,"context_line":""},{"line_number":563,"context_line":"Deleting a port that is an active parent in a trunk is forbidden. If the parent port has"},{"line_number":564,"context_line":"no trunk associated, it can be deleted. The OVS agent doesn\u0027t need to perform any action."},{"line_number":565,"context_line":""},{"line_number":566,"context_line":"Trunk port deletion"},{"line_number":567,"context_line":"+++++++++++++++++++"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_988261c7","line":564,"range":{"start_line":563,"start_character":66,"end_line":564,"end_character":39},"in_reply_to":"7aa08908_7457e325","updated":"2016-06-09 19:06:35.000000000","message":"I think this means \"if the parent port has no trunk associated, then it\u0027s a regular port at that point and can be deleted as such\".","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"c786d2f73a8931a2a34bf5b950122e25fec95064","unresolved":false,"context_lines":[{"line_number":560,"context_line":"Parent port deletion"},{"line_number":561,"context_line":"++++++++++++++++++++"},{"line_number":562,"context_line":""},{"line_number":563,"context_line":"Deleting a port that is an active parent in a trunk is forbidden. If the parent port has"},{"line_number":564,"context_line":"no trunk associated, it can be deleted. The OVS agent doesn\u0027t need to perform any action."},{"line_number":565,"context_line":""},{"line_number":566,"context_line":"Trunk port deletion"},{"line_number":567,"context_line":"+++++++++++++++++++"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_4620ef9b","line":564,"range":{"start_line":563,"start_character":66,"end_line":564,"end_character":39},"in_reply_to":"7aa08908_988261c7","updated":"2016-06-10 12:29:36.000000000","message":"Got it, thanks.","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"5f2a48a56469f2a042ead7b55cbfe6ff908d9db1","unresolved":false,"context_lines":[{"line_number":563,"context_line":"Deleting a port that is an active parent in a trunk is forbidden. If the parent port has"},{"line_number":564,"context_line":"no trunk associated, it can be deleted. The OVS agent doesn\u0027t need to perform any action."},{"line_number":565,"context_line":""},{"line_number":566,"context_line":"Trunk port deletion"},{"line_number":567,"context_line":"+++++++++++++++++++"},{"line_number":568,"context_line":""},{"line_number":569,"context_line":"When Nova deletes a VM, it doesn\u0027t delete the corresponding Neutron port only if it was"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_3428bb99","line":566,"range":{"start_line":566,"start_character":0,"end_line":566,"end_character":10},"updated":"2016-06-09 10:01:48.000000000","message":"s/Trunk port/Trunk/","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"5538f5b25cc28c31e568eadea620fd221987b714","unresolved":false,"context_lines":[{"line_number":563,"context_line":"Deleting a port that is an active parent in a trunk is forbidden. If the parent port has"},{"line_number":564,"context_line":"no trunk associated, it can be deleted. The OVS agent doesn\u0027t need to perform any action."},{"line_number":565,"context_line":""},{"line_number":566,"context_line":"Trunk port deletion"},{"line_number":567,"context_line":"+++++++++++++++++++"},{"line_number":568,"context_line":""},{"line_number":569,"context_line":"When Nova deletes a VM, it doesn\u0027t delete the corresponding Neutron port only if it was"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_d37a58b9","line":566,"range":{"start_line":566,"start_character":0,"end_line":566,"end_character":10},"in_reply_to":"7aa08908_3428bb99","updated":"2016-06-14 13:36:11.000000000","message":"Done","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"5f2a48a56469f2a042ead7b55cbfe6ff908d9db1","unresolved":false,"context_lines":[{"line_number":570,"context_line":"created when booting the VM. In the vlan-aware-vm case the parent port is passed to Nova, so"},{"line_number":571,"context_line":"the port data will remain in the DB after the VM deletion. Nova will delete"},{"line_number":572,"context_line":"the trunk bridge associated with the trunk port in unplug(). The L2 agent"},{"line_number":573,"context_line":"will detect that and notify the Neutron server that the parent port is down."},{"line_number":574,"context_line":""},{"line_number":575,"context_line":"When a trunk port that it\u0027s used by a VM is deleted through the API, Neutron server will"},{"line_number":576,"context_line":"send a network-vif-delete event to Nova. Nova will unplug the vif of the VM from the"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_6e542fca","line":573,"range":{"start_line":573,"start_character":0,"end_line":573,"end_character":16},"updated":"2016-06-09 10:01:48.000000000","message":"When nova/vif-driver deletes the bridge the agent could detect the deletion of the trunk tap ovs interface instead of the bridge deletion. Then again the monitor logic can be kept simpler.","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":4187,"name":"Ryan Tidwell","email":"rtidwell@suse.com","username":"ryan-tidwell"},"change_message_id":"cbb7a818b0f3765d6070da1ee7911d942507c6b7","unresolved":false,"context_lines":[{"line_number":570,"context_line":"created when booting the VM. In the vlan-aware-vm case the parent port is passed to Nova, so"},{"line_number":571,"context_line":"the port data will remain in the DB after the VM deletion. Nova will delete"},{"line_number":572,"context_line":"the trunk bridge associated with the trunk port in unplug(). The L2 agent"},{"line_number":573,"context_line":"will detect that and notify the Neutron server that the parent port is down."},{"line_number":574,"context_line":""},{"line_number":575,"context_line":"When a trunk port that it\u0027s used by a VM is deleted through the API, Neutron server will"},{"line_number":576,"context_line":"send a network-vif-delete event to Nova. Nova will unplug the vif of the VM from the"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_2f3ac834","line":573,"range":{"start_line":573,"start_character":0,"end_line":573,"end_character":16},"in_reply_to":"7aa08908_6e542fca","updated":"2016-06-10 17:40:19.000000000","message":"As you suggested monitoring br-int for the trunk tap keeps the monitoring on br-int, which I think means we can be less invasive.","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"5f2a48a56469f2a042ead7b55cbfe6ff908d9db1","unresolved":false,"context_lines":[{"line_number":572,"context_line":"the trunk bridge associated with the trunk port in unplug(). The L2 agent"},{"line_number":573,"context_line":"will detect that and notify the Neutron server that the parent port is down."},{"line_number":574,"context_line":""},{"line_number":575,"context_line":"When a trunk port that it\u0027s used by a VM is deleted through the API, Neutron server will"},{"line_number":576,"context_line":"send a network-vif-delete event to Nova. Nova will unplug the vif of the VM from the"},{"line_number":577,"context_line":"trunk bridge and delete it."},{"line_number":578,"context_line":""},{"line_number":579,"context_line":"If the trunk port has subports they should be deleted too. The OVS agent will take care of deleting"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_2e1fc70a","line":576,"range":{"start_line":575,"start_character":69,"end_line":576,"end_character":39},"updated":"2016-06-09 10:01:48.000000000","message":"Just as above in line 446.","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":7448,"name":"Carl Baldwin","email":"carl@ecbaldwin.net","username":"carl-baldwin"},"change_message_id":"ad5cc759a71392ee4185b91753fe954cf5ab94da","unresolved":false,"context_lines":[{"line_number":573,"context_line":"will detect that and notify the Neutron server that the parent port is down."},{"line_number":574,"context_line":""},{"line_number":575,"context_line":"When a trunk port that it\u0027s used by a VM is deleted through the API, Neutron server will"},{"line_number":576,"context_line":"send a network-vif-delete event to Nova. Nova will unplug the vif of the VM from the"},{"line_number":577,"context_line":"trunk bridge and delete it."},{"line_number":578,"context_line":""},{"line_number":579,"context_line":"If the trunk port has subports they should be deleted too. The OVS agent will take care of deleting"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_7405b9c0","line":576,"in_reply_to":"7aa08908_2e1fc70a","updated":"2016-06-10 22:10:01.000000000","message":"@Bence:  Also see my reply there.","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"5f2a48a56469f2a042ead7b55cbfe6ff908d9db1","unresolved":false,"context_lines":[{"line_number":583,"context_line":"Subport deletion"},{"line_number":584,"context_line":"++++++++++++++++"},{"line_number":585,"context_line":""},{"line_number":586,"context_line":"The OVS agent will delete the patch port corresponding to the subport deleted."},{"line_number":587,"context_line":""},{"line_number":588,"context_line":"Agent resync"},{"line_number":589,"context_line":"~~~~~~~~~~~~"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_aeb8d7e9","line":586,"range":{"start_line":586,"start_character":30,"end_line":586,"end_character":40},"updated":"2016-06-09 10:01:48.000000000","message":"patch cable or patch portS","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"185dbe523a2e243a01e3aa5cf06a93a9a0efa175","unresolved":false,"context_lines":[{"line_number":583,"context_line":"Subport deletion"},{"line_number":584,"context_line":"++++++++++++++++"},{"line_number":585,"context_line":""},{"line_number":586,"context_line":"The OVS agent will delete the patch port corresponding to the subport deleted."},{"line_number":587,"context_line":""},{"line_number":588,"context_line":"Agent resync"},{"line_number":589,"context_line":"~~~~~~~~~~~~"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_f8a60552","line":586,"range":{"start_line":586,"start_character":30,"end_line":586,"end_character":40},"in_reply_to":"7aa08908_aeb8d7e9","updated":"2016-06-09 19:06:35.000000000","message":"patch port pair?","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"5538f5b25cc28c31e568eadea620fd221987b714","unresolved":false,"context_lines":[{"line_number":583,"context_line":"Subport deletion"},{"line_number":584,"context_line":"++++++++++++++++"},{"line_number":585,"context_line":""},{"line_number":586,"context_line":"The OVS agent will delete the patch port corresponding to the subport deleted."},{"line_number":587,"context_line":""},{"line_number":588,"context_line":"Agent resync"},{"line_number":589,"context_line":"~~~~~~~~~~~~"}],"source_content_type":"text/x-rst","patch_set":4,"id":"7aa08908_1e2b4783","line":586,"range":{"start_line":586,"start_character":30,"end_line":586,"end_character":40},"in_reply_to":"7aa08908_f8a60552","updated":"2016-06-14 13:36:11.000000000","message":"Done","commit_id":"38226a126e05fc22281a39052961f2328dd5be8e"},{"author":{"_account_id":21273,"name":"Ilya Chukhnakov","email":"Ilya.chukhnakov@gmail.com","username":"ichukhnakov"},"change_message_id":"7d43d54daaa23ece6ac62ec3fedf565c36a6528f","unresolved":false,"context_lines":[{"line_number":304,"context_line":"Implementation VLAN Interfaces (Option A)"},{"line_number":305,"context_line":"~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~"},{"line_number":306,"context_line":"This implementation doesn\u0027t require any modification of the vif-drivers since"},{"line_number":307,"context_line":"Nova will plug the vif of the VM the same way as it does for traditional ports."},{"line_number":308,"context_line":""},{"line_number":309,"context_line":"Trunk port creation"},{"line_number":310,"context_line":"+++++++++++++++++++"}],"source_content_type":"text/x-rst","patch_set":5,"id":"7aa08908_fa83d253","line":307,"range":{"start_line":307,"start_character":0,"end_line":307,"end_character":32},"updated":"2016-06-14 19:52:06.000000000","message":"We need to clarify here that Nova will only plug the trunk port, but sub-port plugging would be dealt with by OVS agent as the following sections suggest.","commit_id":"c6d9cfab80d6b917cab1b2449e99cbf53eba1a50"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"fa57e9dd1df41ee2ced49fe18631a30747f05038","unresolved":false,"context_lines":[{"line_number":304,"context_line":"Implementation VLAN Interfaces (Option A)"},{"line_number":305,"context_line":"~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~"},{"line_number":306,"context_line":"This implementation doesn\u0027t require any modification of the vif-drivers since"},{"line_number":307,"context_line":"Nova will plug the vif of the VM the same way as it does for traditional ports."},{"line_number":308,"context_line":""},{"line_number":309,"context_line":"Trunk port creation"},{"line_number":310,"context_line":"+++++++++++++++++++"}],"source_content_type":"text/x-rst","patch_set":5,"id":"7aa08908_481e830e","line":307,"range":{"start_line":307,"start_character":0,"end_line":307,"end_character":32},"in_reply_to":"7aa08908_fa83d253","updated":"2016-06-15 09:31:12.000000000","message":"Technically Nova doesn\u0027t plug any port, it plugs a vif. Anyway as you say it\u0027s explained in the next sections, I don\u0027t want to repeat the same info.","commit_id":"c6d9cfab80d6b917cab1b2449e99cbf53eba1a50"},{"author":{"_account_id":21273,"name":"Ilya Chukhnakov","email":"Ilya.chukhnakov@gmail.com","username":"ichukhnakov"},"change_message_id":"7d43d54daaa23ece6ac62ec3fedf565c36a6528f","unresolved":false,"context_lines":[{"line_number":309,"context_line":"Trunk port creation"},{"line_number":310,"context_line":"+++++++++++++++++++"},{"line_number":311,"context_line":"A VM is spawned passing to Nova the port-id of a parent port associated with"},{"line_number":312,"context_line":"a trunk. Nova will create the tap interface and will plug it into"},{"line_number":313,"context_line":"br-int or into the firewall bridge if using iptables firewall. In the"},{"line_number":314,"context_line":"external-ids of the port Nova will store the port ID of the parent port."},{"line_number":315,"context_line":"The OVS agent detects that a new vif has been plugged. It gets"}],"source_content_type":"text/x-rst","patch_set":5,"id":"7aa08908_3a80baf6","line":312,"range":{"start_line":312,"start_character":9,"end_line":312,"end_character":33},"updated":"2016-06-14 19:52:06.000000000","message":"AFAIK the tap interface is created by libvirt but not directly by Nova. Could we clarify that here please (like \"Nova (libvirt)\")?","commit_id":"c6d9cfab80d6b917cab1b2449e99cbf53eba1a50"},{"author":{"_account_id":21273,"name":"Ilya Chukhnakov","email":"Ilya.chukhnakov@gmail.com","username":"ichukhnakov"},"change_message_id":"7d43d54daaa23ece6ac62ec3fedf565c36a6528f","unresolved":false,"context_lines":[{"line_number":323,"context_line":"If the parent port is associated with one or more subports the agent will process"},{"line_number":324,"context_line":"them as described in the next paragraph."},{"line_number":325,"context_line":""},{"line_number":326,"context_line":"Subport creation"},{"line_number":327,"context_line":"++++++++++++++++"},{"line_number":328,"context_line":"If a subport is added to a parent port but no VM was booted using that parent port"},{"line_number":329,"context_line":"yet, no L2 agent will process it (because at that point the parent port is"}],"source_content_type":"text/x-rst","patch_set":5,"id":"7aa08908_f03bf199","line":326,"range":{"start_line":326,"start_character":0,"end_line":326,"end_character":16},"updated":"2016-06-14 19:52:06.000000000","message":"Could we state here explicitly if we do or do not signal Nova about plugging sub-ports as in \"Trunk port creation\" section?","commit_id":"c6d9cfab80d6b917cab1b2449e99cbf53eba1a50"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"fa57e9dd1df41ee2ced49fe18631a30747f05038","unresolved":false,"context_lines":[{"line_number":323,"context_line":"If the parent port is associated with one or more subports the agent will process"},{"line_number":324,"context_line":"them as described in the next paragraph."},{"line_number":325,"context_line":""},{"line_number":326,"context_line":"Subport creation"},{"line_number":327,"context_line":"++++++++++++++++"},{"line_number":328,"context_line":"If a subport is added to a parent port but no VM was booted using that parent port"},{"line_number":329,"context_line":"yet, no L2 agent will process it (because at that point the parent port is"}],"source_content_type":"text/x-rst","patch_set":5,"id":"7aa08908_889d0b69","line":326,"range":{"start_line":326,"start_character":0,"end_line":326,"end_character":16},"in_reply_to":"7aa08908_f03bf199","updated":"2016-06-15 09:31:12.000000000","message":"We don\u0027t leak details to Nova. Nova is unaware of a port being associated to a trunk. Subports are managed only by Neutron, hence there\u0027s no notification sent to Nova about that.","commit_id":"c6d9cfab80d6b917cab1b2449e99cbf53eba1a50"},{"author":{"_account_id":21273,"name":"Ilya Chukhnakov","email":"Ilya.chukhnakov@gmail.com","username":"ichukhnakov"},"change_message_id":"7d43d54daaa23ece6ac62ec3fedf565c36a6528f","unresolved":false,"context_lines":[{"line_number":325,"context_line":""},{"line_number":326,"context_line":"Subport creation"},{"line_number":327,"context_line":"++++++++++++++++"},{"line_number":328,"context_line":"If a subport is added to a parent port but no VM was booted using that parent port"},{"line_number":329,"context_line":"yet, no L2 agent will process it (because at that point the parent port is"},{"line_number":330,"context_line":"not bound to any host))."},{"line_number":331,"context_line":"When a subport is created for a parent port and a VM that uses that parent port is"},{"line_number":332,"context_line":"already running, the OVS agent will create a VLAN interface on the VM tap"},{"line_number":333,"context_line":"using the VLAN ID specified in the subport segmentation id."}],"source_content_type":"text/x-rst","patch_set":5,"id":"7aa08908_90fb8dc0","line":330,"range":{"start_line":328,"start_character":0,"end_line":330,"end_character":24},"updated":"2016-06-14 19:52:06.000000000","message":"How do we handle the situation when subport is added while the VM is being booted but the parent port is not bound yet?","commit_id":"c6d9cfab80d6b917cab1b2449e99cbf53eba1a50"},{"author":{"_account_id":21273,"name":"Ilya Chukhnakov","email":"Ilya.chukhnakov@gmail.com","username":"ichukhnakov"},"change_message_id":"58743a65b683b0923a6c70ef2a554e8b52a872fe","unresolved":false,"context_lines":[{"line_number":325,"context_line":""},{"line_number":326,"context_line":"Subport creation"},{"line_number":327,"context_line":"++++++++++++++++"},{"line_number":328,"context_line":"If a subport is added to a parent port but no VM was booted using that parent port"},{"line_number":329,"context_line":"yet, no L2 agent will process it (because at that point the parent port is"},{"line_number":330,"context_line":"not bound to any host))."},{"line_number":331,"context_line":"When a subport is created for a parent port and a VM that uses that parent port is"},{"line_number":332,"context_line":"already running, the OVS agent will create a VLAN interface on the VM tap"},{"line_number":333,"context_line":"using the VLAN ID specified in the subport segmentation id."}],"source_content_type":"text/x-rst","patch_set":5,"id":"7aa08908_68fb95c1","line":330,"range":{"start_line":328,"start_character":0,"end_line":330,"end_character":24},"in_reply_to":"7aa08908_2b84ead0","updated":"2016-06-14 20:58:28.000000000","message":"@Rawlin, thats true, but what I meant is that I think this document needs some implementation details about how RPC/agent extension are involved. Sorry for poorly formulating my thoughts :/","commit_id":"c6d9cfab80d6b917cab1b2449e99cbf53eba1a50"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"fa57e9dd1df41ee2ced49fe18631a30747f05038","unresolved":false,"context_lines":[{"line_number":325,"context_line":""},{"line_number":326,"context_line":"Subport creation"},{"line_number":327,"context_line":"++++++++++++++++"},{"line_number":328,"context_line":"If a subport is added to a parent port but no VM was booted using that parent port"},{"line_number":329,"context_line":"yet, no L2 agent will process it (because at that point the parent port is"},{"line_number":330,"context_line":"not bound to any host))."},{"line_number":331,"context_line":"When a subport is created for a parent port and a VM that uses that parent port is"},{"line_number":332,"context_line":"already running, the OVS agent will create a VLAN interface on the VM tap"},{"line_number":333,"context_line":"using the VLAN ID specified in the subport segmentation id."}],"source_content_type":"text/x-rst","patch_set":5,"id":"7aa08908_c2c41210","line":330,"range":{"start_line":328,"start_character":0,"end_line":330,"end_character":24},"in_reply_to":"7aa08908_68fb95c1","updated":"2016-06-15 09:31:12.000000000","message":"@Ilya as stated in the comments of PS1, RPC is left out from this document, it will be tackled in another one","commit_id":"c6d9cfab80d6b917cab1b2449e99cbf53eba1a50"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"6cfe2d52dd1a62bcc85c24ed0d1b596a150a50a5","unresolved":false,"context_lines":[{"line_number":325,"context_line":""},{"line_number":326,"context_line":"Subport creation"},{"line_number":327,"context_line":"++++++++++++++++"},{"line_number":328,"context_line":"If a subport is added to a parent port but no VM was booted using that parent port"},{"line_number":329,"context_line":"yet, no L2 agent will process it (because at that point the parent port is"},{"line_number":330,"context_line":"not bound to any host))."},{"line_number":331,"context_line":"When a subport is created for a parent port and a VM that uses that parent port is"},{"line_number":332,"context_line":"already running, the OVS agent will create a VLAN interface on the VM tap"},{"line_number":333,"context_line":"using the VLAN ID specified in the subport segmentation id."}],"source_content_type":"text/x-rst","patch_set":5,"id":"7aa08908_2b84ead0","line":330,"range":{"start_line":328,"start_character":0,"end_line":330,"end_character":24},"in_reply_to":"7aa08908_90fb8dc0","updated":"2016-06-14 20:18:05.000000000","message":"Subports will be bound at the same time the trunk\u0027s parent port is bound. If a subport is added concurrently before the trunk\u0027s parent port is bound, then it will become bound when the trunk\u0027s parent port is bound. Subports added to a trunk with a bound parent port should be considered as \"subport created\" events that should be pushed to the agent.","commit_id":"c6d9cfab80d6b917cab1b2449e99cbf53eba1a50"},{"author":{"_account_id":21273,"name":"Ilya Chukhnakov","email":"Ilya.chukhnakov@gmail.com","username":"ichukhnakov"},"change_message_id":"7d43d54daaa23ece6ac62ec3fedf565c36a6528f","unresolved":false,"context_lines":[{"line_number":363,"context_line":""},{"line_number":364,"context_line":"Let\u0027s assume the subport is on network2 and uses segmentation ID 100."},{"line_number":365,"context_line":"In the case of hybrid plugging the OVS agent will have to create the firewall"},{"line_number":366,"context_line":"bridge (qbr2), create tap1.100 and plug it into qbr2. It will connect qbr2 to"},{"line_number":367,"context_line":"br-int and set the subport ID in the external-ids of port 2."},{"line_number":368,"context_line":""},{"line_number":369,"context_line":"*Inbound traffic from the VM point of view*"}],"source_content_type":"text/x-rst","patch_set":5,"id":"7aa08908_5a10264e","line":366,"range":{"start_line":366,"start_character":15,"end_line":366,"end_character":30},"updated":"2016-06-14 19:52:06.000000000","message":"How do we make sure that \u0027tap1\u0027 exists? It is created by libvirt after the linux bridge is plugged into OVS. So there is potentially a race condition that by the time OVS agent wants to create \u0027tap1.100\u0027 the \u0027tap1\u0027 device has not been created yet.","commit_id":"c6d9cfab80d6b917cab1b2449e99cbf53eba1a50"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"fa57e9dd1df41ee2ced49fe18631a30747f05038","unresolved":false,"context_lines":[{"line_number":363,"context_line":""},{"line_number":364,"context_line":"Let\u0027s assume the subport is on network2 and uses segmentation ID 100."},{"line_number":365,"context_line":"In the case of hybrid plugging the OVS agent will have to create the firewall"},{"line_number":366,"context_line":"bridge (qbr2), create tap1.100 and plug it into qbr2. It will connect qbr2 to"},{"line_number":367,"context_line":"br-int and set the subport ID in the external-ids of port 2."},{"line_number":368,"context_line":""},{"line_number":369,"context_line":"*Inbound traffic from the VM point of view*"}],"source_content_type":"text/x-rst","patch_set":5,"id":"7aa08908_e2eaee73","line":366,"range":{"start_line":366,"start_character":15,"end_line":366,"end_character":30},"in_reply_to":"7aa08908_5a10264e","updated":"2016-06-15 09:31:12.000000000","message":"as stated at line 328 the subport creation when the parent port is not bound to any VM is a no op for the agent. So tap1 must exists at this point.","commit_id":"c6d9cfab80d6b917cab1b2449e99cbf53eba1a50"},{"author":{"_account_id":21273,"name":"Ilya Chukhnakov","email":"Ilya.chukhnakov@gmail.com","username":"ichukhnakov"},"change_message_id":"2ed172e377c9d675992eae644bb7bb48b904934e","unresolved":false,"context_lines":[{"line_number":363,"context_line":""},{"line_number":364,"context_line":"Let\u0027s assume the subport is on network2 and uses segmentation ID 100."},{"line_number":365,"context_line":"In the case of hybrid plugging the OVS agent will have to create the firewall"},{"line_number":366,"context_line":"bridge (qbr2), create tap1.100 and plug it into qbr2. It will connect qbr2 to"},{"line_number":367,"context_line":"br-int and set the subport ID in the external-ids of port 2."},{"line_number":368,"context_line":""},{"line_number":369,"context_line":"*Inbound traffic from the VM point of view*"}],"source_content_type":"text/x-rst","patch_set":5,"id":"7aa08908_ab91d3f3","line":366,"range":{"start_line":366,"start_character":15,"end_line":366,"end_character":30},"in_reply_to":"7aa08908_902fe687","updated":"2016-06-15 14:01:58.000000000","message":"Most races are unlikely to happen yet they do. I suggest that this document should at least describe the problem.\n\nSo if the sub-port creation does fail what would be the outcome? Will the trunk port creation fail and result in VM failing to boot or will the VM boot with that sub-port network missing? I consider either of those as bad UX. Maybe we could provide some recovery mechanism?","commit_id":"c6d9cfab80d6b917cab1b2449e99cbf53eba1a50"},{"author":{"_account_id":21273,"name":"Ilya Chukhnakov","email":"Ilya.chukhnakov@gmail.com","username":"ichukhnakov"},"change_message_id":"c16b927450ab1a33fdc61e68e08328a0e0ae90a7","unresolved":false,"context_lines":[{"line_number":363,"context_line":""},{"line_number":364,"context_line":"Let\u0027s assume the subport is on network2 and uses segmentation ID 100."},{"line_number":365,"context_line":"In the case of hybrid plugging the OVS agent will have to create the firewall"},{"line_number":366,"context_line":"bridge (qbr2), create tap1.100 and plug it into qbr2. It will connect qbr2 to"},{"line_number":367,"context_line":"br-int and set the subport ID in the external-ids of port 2."},{"line_number":368,"context_line":""},{"line_number":369,"context_line":"*Inbound traffic from the VM point of view*"}],"source_content_type":"text/x-rst","patch_set":5,"id":"7aa08908_fc928bfb","line":366,"range":{"start_line":366,"start_character":15,"end_line":366,"end_character":30},"in_reply_to":"7aa08908_e2eaee73","updated":"2016-06-15 12:43:13.000000000","message":"@Rossella, please take a look at Nova\u0027s _create_domain_and_network [1] (called by \u0027spawn\u0027 in _build_and_run_instance [2]). First it plugs vifs [3] and then calls libvirt to create the VM [4]. In-between those two calls the \u0027tap\u0027 does not exist, but the linux bridge does and veth is already plugged into OVS bridge [5]. Also note that _build_resources [2] invokes \u0027allocate_for_instance\u0027 [6] that updates the port with bindings (it seems that call is asynchronous).\n\nSo as far as I understand, when OVS Agent detects the trunk port (veth) the port could have been bound already, but the \u0027tap\u0027 device is not guaranteed to exist yet. I believe that\u0027s similar to \u0027coordinated assumptions\u0027 problem that Kevin described in [7].\n\n[1] https://github.com/openstack/nova/blob/b04da5439daf7733c09d6a2090890dd83e59b5e1/nova/virt/libvirt/driver.py#L4675\n[2] https://github.com/openstack/nova/blob/b04da5439daf7733c09d6a2090890dd83e59b5e1/nova/compute/manager.py#L1871-L1892\n[3] https://github.com/openstack/nova/blob/b04da5439daf7733c09d6a2090890dd83e59b5e1/nova/virt/libvirt/driver.py#L4712\n[4] https://github.com/openstack/nova/blob/b04da5439daf7733c09d6a2090890dd83e59b5e1/nova/virt/libvirt/driver.py#L4719\n[5] https://github.com/openstack/nova/blob/b04da5439daf7733c09d6a2090890dd83e59b5e1/nova/virt/libvirt/vif.py#L508\n[6] https://github.com/openstack/nova/blob/b04da5439daf7733c09d6a2090890dd83e59b5e1/nova/network/neutronv2/api.py#L483\n[7] http://lists.openstack.org/pipermail/openstack-dev/2016-June/097384.html","commit_id":"c6d9cfab80d6b917cab1b2449e99cbf53eba1a50"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"3e574e0a5b05c59ca377881f2089fb06b306cace","unresolved":false,"context_lines":[{"line_number":363,"context_line":""},{"line_number":364,"context_line":"Let\u0027s assume the subport is on network2 and uses segmentation ID 100."},{"line_number":365,"context_line":"In the case of hybrid plugging the OVS agent will have to create the firewall"},{"line_number":366,"context_line":"bridge (qbr2), create tap1.100 and plug it into qbr2. It will connect qbr2 to"},{"line_number":367,"context_line":"br-int and set the subport ID in the external-ids of port 2."},{"line_number":368,"context_line":""},{"line_number":369,"context_line":"*Inbound traffic from the VM point of view*"}],"source_content_type":"text/x-rst","patch_set":5,"id":"7aa08908_902fe687","line":366,"range":{"start_line":366,"start_character":15,"end_line":366,"end_character":30},"in_reply_to":"7aa08908_fc928bfb","updated":"2016-06-15 13:31:05.000000000","message":"@Ilya between point [3] and [4] the tap might not exists. Anyway when the agent detects a new port, it issues a call to Neutron server to get the device details, otherwise it cannot process it. So there\u0027s enough time for libvirt to create the vif. Yes there\u0027s a very small chance of a race. In that case the subport creation will fail.","commit_id":"c6d9cfab80d6b917cab1b2449e99cbf53eba1a50"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"a16260818d4f0f6d1de0a834b118ddd612c5f694","unresolved":false,"context_lines":[{"line_number":444,"context_line":"will detect that deletion and notify the Neutron server that the parent port is down."},{"line_number":445,"context_line":"The OVS agent will clean up the corresponding subports as explained in the next paragraph."},{"line_number":446,"context_line":""},{"line_number":447,"context_line":"The deletion of a trunk that it\u0027s used by a VM is not allowed."},{"line_number":448,"context_line":""},{"line_number":449,"context_line":"Subport deletion"},{"line_number":450,"context_line":"++++++++++++++++"}],"source_content_type":"text/x-rst","patch_set":5,"id":"7aa08908_c41efbb9","line":447,"range":{"start_line":447,"start_character":0,"end_line":447,"end_character":62},"updated":"2016-06-14 20:09:47.000000000","message":"This sentence still isn\u0027t very clear.\n\nWhich of the following are we going to enforce?\n1) the trunk can be deleted (leaving the parent port intact)\n2) the trunk cannot be deleted (it will be deleted when it\u0027s parent port is deleted)","commit_id":"c6d9cfab80d6b917cab1b2449e99cbf53eba1a50"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"fa57e9dd1df41ee2ced49fe18631a30747f05038","unresolved":false,"context_lines":[{"line_number":444,"context_line":"will detect that deletion and notify the Neutron server that the parent port is down."},{"line_number":445,"context_line":"The OVS agent will clean up the corresponding subports as explained in the next paragraph."},{"line_number":446,"context_line":""},{"line_number":447,"context_line":"The deletion of a trunk that it\u0027s used by a VM is not allowed."},{"line_number":448,"context_line":""},{"line_number":449,"context_line":"Subport deletion"},{"line_number":450,"context_line":"++++++++++++++++"}],"source_content_type":"text/x-rst","patch_set":5,"id":"7aa08908_c80f138a","line":447,"range":{"start_line":447,"start_character":0,"end_line":447,"end_character":62},"in_reply_to":"7aa08908_c41efbb9","updated":"2016-06-15 09:31:12.000000000","message":"Thanks, I will clarify","commit_id":"c6d9cfab80d6b917cab1b2449e99cbf53eba1a50"},{"author":{"_account_id":21273,"name":"Ilya Chukhnakov","email":"Ilya.chukhnakov@gmail.com","username":"ichukhnakov"},"change_message_id":"7d43d54daaa23ece6ac62ec3fedf565c36a6528f","unresolved":false,"context_lines":[{"line_number":452,"context_line":"Removing a subport that is associated with a parent port that was not used to boot any"},{"line_number":453,"context_line":"VM is a no op from the OVS agent perspective."},{"line_number":454,"context_line":"When a subport associated with a parent port that was used to boot a VM is deleted,"},{"line_number":455,"context_line":"the OVS agent will take care of removing the VLAN interface, the firewall bridge"},{"line_number":456,"context_line":"if using iptables firewall and the port on br-int."},{"line_number":457,"context_line":""},{"line_number":458,"context_line":""}],"source_content_type":"text/x-rst","patch_set":5,"id":"7aa08908_b5eafbd5","line":455,"range":{"start_line":455,"start_character":32,"end_line":455,"end_character":59},"updated":"2016-06-14 19:52:06.000000000","message":"It is removed automatically when the parent \u0027tap\u0027 interface is removed, so OVS agent won\u0027t be able to remove it.","commit_id":"c6d9cfab80d6b917cab1b2449e99cbf53eba1a50"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"fa57e9dd1df41ee2ced49fe18631a30747f05038","unresolved":false,"context_lines":[{"line_number":452,"context_line":"Removing a subport that is associated with a parent port that was not used to boot any"},{"line_number":453,"context_line":"VM is a no op from the OVS agent perspective."},{"line_number":454,"context_line":"When a subport associated with a parent port that was used to boot a VM is deleted,"},{"line_number":455,"context_line":"the OVS agent will take care of removing the VLAN interface, the firewall bridge"},{"line_number":456,"context_line":"if using iptables firewall and the port on br-int."},{"line_number":457,"context_line":""},{"line_number":458,"context_line":""}],"source_content_type":"text/x-rst","patch_set":5,"id":"7aa08908_e843ef11","line":455,"range":{"start_line":455,"start_character":32,"end_line":455,"end_character":59},"in_reply_to":"7aa08908_b5eafbd5","updated":"2016-06-15 09:31:12.000000000","message":"Done","commit_id":"c6d9cfab80d6b917cab1b2449e99cbf53eba1a50"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"a16260818d4f0f6d1de0a834b118ddd612c5f694","unresolved":false,"context_lines":[{"line_number":568,"context_line":"the trunk bridge associated with the trunk in unplug(). The L2 agent"},{"line_number":569,"context_line":"will detect that and notify the Neutron server that the parent port is down."},{"line_number":570,"context_line":""},{"line_number":571,"context_line":"The deletion of a trunk that it\u0027s used by a VM is not allowed."},{"line_number":572,"context_line":""},{"line_number":573,"context_line":"Subport deletion"},{"line_number":574,"context_line":"++++++++++++++++"}],"source_content_type":"text/x-rst","patch_set":5,"id":"7aa08908_77225b53","line":571,"range":{"start_line":571,"start_character":0,"end_line":571,"end_character":62},"updated":"2016-06-14 20:09:47.000000000","message":"See comment on L447","commit_id":"c6d9cfab80d6b917cab1b2449e99cbf53eba1a50"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"fa57e9dd1df41ee2ced49fe18631a30747f05038","unresolved":false,"context_lines":[{"line_number":568,"context_line":"the trunk bridge associated with the trunk in unplug(). The L2 agent"},{"line_number":569,"context_line":"will detect that and notify the Neutron server that the parent port is down."},{"line_number":570,"context_line":""},{"line_number":571,"context_line":"The deletion of a trunk that it\u0027s used by a VM is not allowed."},{"line_number":572,"context_line":""},{"line_number":573,"context_line":"Subport deletion"},{"line_number":574,"context_line":"++++++++++++++++"}],"source_content_type":"text/x-rst","patch_set":5,"id":"7aa08908_635d6015","line":571,"range":{"start_line":571,"start_character":0,"end_line":571,"end_character":62},"in_reply_to":"7aa08908_77225b53","updated":"2016-06-15 09:31:12.000000000","message":"Done","commit_id":"c6d9cfab80d6b917cab1b2449e99cbf53eba1a50"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"363cac946e312470c11058952c3561bb08462dc4","unresolved":false,"context_lines":[{"line_number":445,"context_line":"The OVS agent will clean up the corresponding subports as explained in the next paragraph."},{"line_number":446,"context_line":""},{"line_number":447,"context_line":"The deletion of a trunk that it\u0027s used by a VM is not allowed."},{"line_number":448,"context_line":"The trunk can be deleted (leaving the parent port intact) when the parent port is not"},{"line_number":449,"context_line":"used by any VM. After the trunk is deleted, the parent port can also be deleted."},{"line_number":450,"context_line":""},{"line_number":451,"context_line":"Subport deletion"},{"line_number":452,"context_line":"++++++++++++++++"}],"source_content_type":"text/x-rst","patch_set":6,"id":"7aa08908_3ceb687a","line":449,"range":{"start_line":448,"start_character":58,"end_line":449,"end_character":14},"updated":"2016-06-15 15:25:51.000000000","message":"So if the parent port is bound, we will return an error when trying to delete a trunk? And if a port is currently a parent port of a trunk, it cannot be deleted.\n\nWe may also want to mention this same set of constraints for subports in the section below. \"A port cannot be deleted if it belongs to a subport. The subport must be deleted first, then its parent port may also be deleted.\"","commit_id":"bb1ad276a7c7f16d1f68e31d8d1c1cdb11056204"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"89ea1bc5ba2cbe20bf3a3314d2a0bf4401e8f2c3","unresolved":false,"context_lines":[{"line_number":445,"context_line":"The OVS agent will clean up the corresponding subports as explained in the next paragraph."},{"line_number":446,"context_line":""},{"line_number":447,"context_line":"The deletion of a trunk that it\u0027s used by a VM is not allowed."},{"line_number":448,"context_line":"The trunk can be deleted (leaving the parent port intact) when the parent port is not"},{"line_number":449,"context_line":"used by any VM. After the trunk is deleted, the parent port can also be deleted."},{"line_number":450,"context_line":""},{"line_number":451,"context_line":"Subport deletion"},{"line_number":452,"context_line":"++++++++++++++++"}],"source_content_type":"text/x-rst","patch_set":6,"id":"3aaa91ec_ba037841","line":449,"range":{"start_line":448,"start_character":58,"end_line":449,"end_character":14},"in_reply_to":"7aa08908_3ceb687a","updated":"2016-06-23 13:40:43.000000000","message":"well if you can\u0027t delete a port associated to a trunk as a consequence you cannot delete a port that has subports, since to have subports you need a trunk. I think this is straightforward, no need to be so verbose","commit_id":"bb1ad276a7c7f16d1f68e31d8d1c1cdb11056204"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"101e70b0c79586d57ecb2a497032233d3ef460f6","unresolved":false,"context_lines":[{"line_number":502,"context_line":"The os-vif driver creates the trunk bridge br-trunk-1 if it does not exist in plug()"},{"line_number":503,"context_line":"with the parent port ID in the external-ids field. It will create the tap interface"},{"line_number":504,"context_line":"tap1 and plug it into br-trunk-1 setting the parent port ID in the external-ids."},{"line_number":505,"context_line":"The OVS agent will be monitoring the creation of new trunk bridges. When it detects"},{"line_number":506,"context_line":"that a new trunk bridge has been created, it will do the following:"},{"line_number":507,"context_line":""},{"line_number":508,"context_line":"ovs-vsctl add-port  br-trunk-1 tp-patch-trunk -- set Interface tp-patch-trunk type\u003dpatch options:peer\u003dtp-patch-int"},{"line_number":509,"context_line":"ovs-vsctl add-port br-int tp-patch-int tag\u003d3 -- set Interface tp-patch-int type\u003dpatch options:peer\u003dtp-patch-trunk"}],"source_content_type":"text/x-rst","patch_set":6,"id":"7aa08908_863937db","line":506,"range":{"start_line":505,"start_character":68,"end_line":506,"end_character":67},"updated":"2016-06-16 12:25:37.000000000","message":"Sorry for coming back to this again, but I still slightly dislike monitoring bridge creation. For regular ports the agent monitors an ovs interface that was created by libvirt. However if we start monitoring the bridge (in comparison to monitoring the trunk tap) that is created by nova, and the trunk tap gets created only later.\n\nAFAICT this does not lead to a race condition, due to nova waiting for both the neutron event and libvirt finishing before resuming the vm, but it still changes the current model. We no longer react to an action of libvirt, but an action of nova. Which can make debugging more confusing and later maintenance more complicated.\n\nI find this kind of variation in the flow of causes and effects unnecessary complexity.","commit_id":"bb1ad276a7c7f16d1f68e31d8d1c1cdb11056204"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"af58efff9e2b4fe38e1094fee5b1eaf0413135f6","unresolved":false,"context_lines":[{"line_number":502,"context_line":"The os-vif driver creates the trunk bridge br-trunk-1 if it does not exist in plug()"},{"line_number":503,"context_line":"with the parent port ID in the external-ids field. It will create the tap interface"},{"line_number":504,"context_line":"tap1 and plug it into br-trunk-1 setting the parent port ID in the external-ids."},{"line_number":505,"context_line":"The OVS agent will be monitoring the creation of new trunk bridges. When it detects"},{"line_number":506,"context_line":"that a new trunk bridge has been created, it will do the following:"},{"line_number":507,"context_line":""},{"line_number":508,"context_line":"ovs-vsctl add-port  br-trunk-1 tp-patch-trunk -- set Interface tp-patch-trunk type\u003dpatch options:peer\u003dtp-patch-int"},{"line_number":509,"context_line":"ovs-vsctl add-port br-int tp-patch-int tag\u003d3 -- set Interface tp-patch-int type\u003dpatch options:peer\u003dtp-patch-trunk"}],"source_content_type":"text/x-rst","patch_set":6,"id":"7aa08908_2e6558e0","line":506,"range":{"start_line":505,"start_character":68,"end_line":506,"end_character":67},"in_reply_to":"7aa08908_0056b905","updated":"2016-06-18 11:37:44.000000000","message":"in the hybrid plug case the ovs agent is already responding to the nova action of adding the port to ovs not libvirt so the behavior is not changing.\n\nthe trunk bridge will have the neutron port uuid in its external ids field but yes the agent should only wire up the trunk when the port is added to the trunk bridge.  the ovs agent does not have to monitor the creation of bridges directly but it does have to monitor the creation of port on bridge other then the br-int.","commit_id":"bb1ad276a7c7f16d1f68e31d8d1c1cdb11056204"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"89ea1bc5ba2cbe20bf3a3314d2a0bf4401e8f2c3","unresolved":false,"context_lines":[{"line_number":502,"context_line":"The os-vif driver creates the trunk bridge br-trunk-1 if it does not exist in plug()"},{"line_number":503,"context_line":"with the parent port ID in the external-ids field. It will create the tap interface"},{"line_number":504,"context_line":"tap1 and plug it into br-trunk-1 setting the parent port ID in the external-ids."},{"line_number":505,"context_line":"The OVS agent will be monitoring the creation of new trunk bridges. When it detects"},{"line_number":506,"context_line":"that a new trunk bridge has been created, it will do the following:"},{"line_number":507,"context_line":""},{"line_number":508,"context_line":"ovs-vsctl add-port  br-trunk-1 tp-patch-trunk -- set Interface tp-patch-trunk type\u003dpatch options:peer\u003dtp-patch-int"},{"line_number":509,"context_line":"ovs-vsctl add-port br-int tp-patch-int tag\u003d3 -- set Interface tp-patch-int type\u003dpatch options:peer\u003dtp-patch-trunk"}],"source_content_type":"text/x-rst","patch_set":6,"id":"3aaa91ec_756f61c5","line":506,"range":{"start_line":505,"start_character":68,"end_line":506,"end_character":67},"in_reply_to":"7aa08908_2e6558e0","updated":"2016-06-23 13:40:43.000000000","message":"I will modify this, the OVS agent will monitor the creation of port on the trunk bridges","commit_id":"bb1ad276a7c7f16d1f68e31d8d1c1cdb11056204"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"8e1c74177e0a158b2272be61b3a65fb397396bbf","unresolved":false,"context_lines":[{"line_number":502,"context_line":"The os-vif driver creates the trunk bridge br-trunk-1 if it does not exist in plug()"},{"line_number":503,"context_line":"with the parent port ID in the external-ids field. It will create the tap interface"},{"line_number":504,"context_line":"tap1 and plug it into br-trunk-1 setting the parent port ID in the external-ids."},{"line_number":505,"context_line":"The OVS agent will be monitoring the creation of new trunk bridges. When it detects"},{"line_number":506,"context_line":"that a new trunk bridge has been created, it will do the following:"},{"line_number":507,"context_line":""},{"line_number":508,"context_line":"ovs-vsctl add-port  br-trunk-1 tp-patch-trunk -- set Interface tp-patch-trunk type\u003dpatch options:peer\u003dtp-patch-int"},{"line_number":509,"context_line":"ovs-vsctl add-port br-int tp-patch-int tag\u003d3 -- set Interface tp-patch-int type\u003dpatch options:peer\u003dtp-patch-trunk"}],"source_content_type":"text/x-rst","patch_set":6,"id":"7aa08908_0056b905","line":506,"range":{"start_line":505,"start_character":68,"end_line":506,"end_character":67},"in_reply_to":"7aa08908_863937db","updated":"2016-06-16 17:07:53.000000000","message":"I agree here. We should be monitoring for a new port getting created on a trunk bridge, rather than just monitoring for a new trunk bridge.\n\nAt the time the bridge is created, the tap might not exist yet, so none of its external_ids will be available either. Without the external_ids, we have no way to get the port_id (it\u0027s found in external_ids:iface-id), unless we somehow use the port_id prefix from the name of the trunk bridge it\u0027s attached to in order to find the full port_id or keep spinning until the tap is present on the bridge. Neither of those approaches sound good to me.","commit_id":"bb1ad276a7c7f16d1f68e31d8d1c1cdb11056204"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"af58efff9e2b4fe38e1094fee5b1eaf0413135f6","unresolved":false,"context_lines":[{"line_number":566,"context_line":""},{"line_number":567,"context_line":"When Nova deletes a VM, it doesn\u0027t delete the corresponding Neutron port only if it was"},{"line_number":568,"context_line":"created when booting the VM. In the vlan-aware-vm case the parent port is passed to Nova, so"},{"line_number":569,"context_line":"the port data will remain in the DB after the VM deletion. Nova will delete"},{"line_number":570,"context_line":"the trunk bridge associated with the trunk in unplug(). The L2 agent"},{"line_number":571,"context_line":"will detect that and notify the Neutron server that the parent port is down."},{"line_number":572,"context_line":""},{"line_number":573,"context_line":"The deletion of a trunk that it\u0027s used by a VM is not allowed."}],"source_content_type":"text/x-rst","patch_set":6,"id":"7aa08908_ae9f88b2","line":570,"range":{"start_line":569,"start_character":59,"end_line":570,"end_character":54},"updated":"2016-06-18 11:37:44.000000000","message":"nova will only delete the port not the bridge. the ovs agent will delete the bridge. after all nova dose not no the br-int is any different from a trunk bridge so it has no way to know it should not delete it.","commit_id":"bb1ad276a7c7f16d1f68e31d8d1c1cdb11056204"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"89ea1bc5ba2cbe20bf3a3314d2a0bf4401e8f2c3","unresolved":false,"context_lines":[{"line_number":566,"context_line":""},{"line_number":567,"context_line":"When Nova deletes a VM, it doesn\u0027t delete the corresponding Neutron port only if it was"},{"line_number":568,"context_line":"created when booting the VM. In the vlan-aware-vm case the parent port is passed to Nova, so"},{"line_number":569,"context_line":"the port data will remain in the DB after the VM deletion. Nova will delete"},{"line_number":570,"context_line":"the trunk bridge associated with the trunk in unplug(). The L2 agent"},{"line_number":571,"context_line":"will detect that and notify the Neutron server that the parent port is down."},{"line_number":572,"context_line":""},{"line_number":573,"context_line":"The deletion of a trunk that it\u0027s used by a VM is not allowed."}],"source_content_type":"text/x-rst","patch_set":6,"id":"3aaa91ec_35831958","line":570,"range":{"start_line":569,"start_character":59,"end_line":570,"end_character":54},"in_reply_to":"5a9d85d2_6e523915","updated":"2016-06-23 13:40:43.000000000","message":"I will change that, Neutron will delete the bridge to simplify the changes needed in os-vif","commit_id":"bb1ad276a7c7f16d1f68e31d8d1c1cdb11056204"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"e69a3ad40711be9249ebbac527383ffd514e92dd","unresolved":false,"context_lines":[{"line_number":566,"context_line":""},{"line_number":567,"context_line":"When Nova deletes a VM, it doesn\u0027t delete the corresponding Neutron port only if it was"},{"line_number":568,"context_line":"created when booting the VM. In the vlan-aware-vm case the parent port is passed to Nova, so"},{"line_number":569,"context_line":"the port data will remain in the DB after the VM deletion. Nova will delete"},{"line_number":570,"context_line":"the trunk bridge associated with the trunk in unplug(). The L2 agent"},{"line_number":571,"context_line":"will detect that and notify the Neutron server that the parent port is down."},{"line_number":572,"context_line":""},{"line_number":573,"context_line":"The deletion of a trunk that it\u0027s used by a VM is not allowed."}],"source_content_type":"text/x-rst","patch_set":6,"id":"5a9d85d2_6e523915","line":570,"range":{"start_line":569,"start_character":59,"end_line":570,"end_character":54},"in_reply_to":"7aa08908_ae9f88b2","updated":"2016-06-20 19:19:09.000000000","message":"I don\u0027t really have a preference either way, but I think whoever is responsible for setting the necessary metadata on the trunk bridge (so that it can be later found and deleted) should be responsible for deleting it.","commit_id":"bb1ad276a7c7f16d1f68e31d8d1c1cdb11056204"},{"author":{"_account_id":15554,"name":"Bence Romsics","email":"bence.romsics@gmail.com","username":"ebenrom","status":"inactive contributor"},"change_message_id":"101e70b0c79586d57ecb2a497032233d3ef460f6","unresolved":false,"context_lines":[{"line_number":570,"context_line":"the trunk bridge associated with the trunk in unplug(). The L2 agent"},{"line_number":571,"context_line":"will detect that and notify the Neutron server that the parent port is down."},{"line_number":572,"context_line":""},{"line_number":573,"context_line":"The deletion of a trunk that it\u0027s used by a VM is not allowed."},{"line_number":574,"context_line":"The trunk can be deleted (leaving the parent port intact) when the parent port is not"},{"line_number":575,"context_line":"used by any VM. After the trunk is deleted, the parent port can also be deleted."},{"line_number":576,"context_line":""}],"source_content_type":"text/x-rst","patch_set":6,"id":"7aa08908_c61fbfb5","line":573,"range":{"start_line":573,"start_character":0,"end_line":573,"end_character":62},"updated":"2016-06-16 12:25:37.000000000","message":"I like the symmetry of this. If a regular port cannot be turned into a trunk port while being used in a vm, a trunk port cannot be turned into a regular port either.","commit_id":"bb1ad276a7c7f16d1f68e31d8d1c1cdb11056204"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"cf613c5e2bd14341ae33fdc7a115a16afa42979a","unresolved":false,"context_lines":[{"line_number":327,"context_line":"++++++++++++++++"},{"line_number":328,"context_line":"If a subport is added to a parent port but no VM was booted using that parent port"},{"line_number":329,"context_line":"yet, no L2 agent will process it (because at that point the parent port is"},{"line_number":330,"context_line":"not bound to any host))."},{"line_number":331,"context_line":"When a subport is created for a parent port and a VM that uses that parent port is"},{"line_number":332,"context_line":"already running, the OVS agent will create a VLAN interface on the VM tap"},{"line_number":333,"context_line":"using the VLAN ID specified in the subport segmentation id. There\u0027s a small possibility"}],"source_content_type":"text/x-rst","patch_set":7,"id":"3aaa91ec_a0d3f75e","line":330,"range":{"start_line":330,"start_character":21,"end_line":330,"end_character":23},"updated":"2016-06-24 20:40:00.000000000","message":"extra paren","commit_id":"94417d5995367aad00912b6d6754d5961134b6a9"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"dd3d1b0a427cc461bd36a7cf5f2744aa89d2ba39","unresolved":false,"context_lines":[{"line_number":327,"context_line":"++++++++++++++++"},{"line_number":328,"context_line":"If a subport is added to a parent port but no VM was booted using that parent port"},{"line_number":329,"context_line":"yet, no L2 agent will process it (because at that point the parent port is"},{"line_number":330,"context_line":"not bound to any host))."},{"line_number":331,"context_line":"When a subport is created for a parent port and a VM that uses that parent port is"},{"line_number":332,"context_line":"already running, the OVS agent will create a VLAN interface on the VM tap"},{"line_number":333,"context_line":"using the VLAN ID specified in the subport segmentation id. There\u0027s a small possibility"}],"source_content_type":"text/x-rst","patch_set":7,"id":"3aaa91ec_4a2966e5","line":330,"range":{"start_line":330,"start_character":21,"end_line":330,"end_character":23},"in_reply_to":"3aaa91ec_a0d3f75e","updated":"2016-06-27 13:52:53.000000000","message":"Done","commit_id":"94417d5995367aad00912b6d6754d5961134b6a9"},{"author":{"_account_id":7448,"name":"Carl Baldwin","email":"carl@ecbaldwin.net","username":"carl-baldwin"},"change_message_id":"ceb85363e740e3dc3601aeafb3f7fe6faed42939","unresolved":false,"context_lines":[{"line_number":332,"context_line":"already running, the OVS agent will create a VLAN interface on the VM tap"},{"line_number":333,"context_line":"using the VLAN ID specified in the subport segmentation id. There\u0027s a small possibility"},{"line_number":334,"context_line":"that a race might occur: the firewall bridge might be created and plugged while the vif"},{"line_number":335,"context_line":"is not there yet. The OVS agent needs to check if the vif exists before trying to create"},{"line_number":336,"context_line":"a subinterface."},{"line_number":337,"context_line":"Let\u0027s see how the models differ when using the iptables firewall or the ovs native"},{"line_number":338,"context_line":"firewall."}],"source_content_type":"text/x-rst","patch_set":7,"id":"3aaa91ec_85bc0dbc","line":335,"range":{"start_line":335,"start_character":32,"end_line":335,"end_character":64},"updated":"2016-06-24 20:54:38.000000000","message":"In the case where it doesn\u0027t exist, I assume it will just give up here.  Then this sub-interface will be added later when L323-L324 above happens.  Is that right?","commit_id":"94417d5995367aad00912b6d6754d5961134b6a9"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"dd3d1b0a427cc461bd36a7cf5f2744aa89d2ba39","unresolved":false,"context_lines":[{"line_number":332,"context_line":"already running, the OVS agent will create a VLAN interface on the VM tap"},{"line_number":333,"context_line":"using the VLAN ID specified in the subport segmentation id. There\u0027s a small possibility"},{"line_number":334,"context_line":"that a race might occur: the firewall bridge might be created and plugged while the vif"},{"line_number":335,"context_line":"is not there yet. The OVS agent needs to check if the vif exists before trying to create"},{"line_number":336,"context_line":"a subinterface."},{"line_number":337,"context_line":"Let\u0027s see how the models differ when using the iptables firewall or the ovs native"},{"line_number":338,"context_line":"firewall."}],"source_content_type":"text/x-rst","patch_set":7,"id":"3aaa91ec_eae93280","line":335,"range":{"start_line":335,"start_character":32,"end_line":335,"end_character":64},"in_reply_to":"3aaa91ec_667e486b","updated":"2016-06-27 13:52:53.000000000","message":"Regardless of the way we implement it (if we use an L2 agent extension or not) there should be some retry mechanism. I won\u0027t add more details regarding that here, if we decide to go with this option we can expand it.","commit_id":"94417d5995367aad00912b6d6754d5961134b6a9"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"34939891858e9076c9af92259f59fc2b61c7bc45","unresolved":false,"context_lines":[{"line_number":332,"context_line":"already running, the OVS agent will create a VLAN interface on the VM tap"},{"line_number":333,"context_line":"using the VLAN ID specified in the subport segmentation id. There\u0027s a small possibility"},{"line_number":334,"context_line":"that a race might occur: the firewall bridge might be created and plugged while the vif"},{"line_number":335,"context_line":"is not there yet. The OVS agent needs to check if the vif exists before trying to create"},{"line_number":336,"context_line":"a subinterface."},{"line_number":337,"context_line":"Let\u0027s see how the models differ when using the iptables firewall or the ovs native"},{"line_number":338,"context_line":"firewall."}],"source_content_type":"text/x-rst","patch_set":7,"id":"3aaa91ec_667e486b","line":335,"range":{"start_line":335,"start_character":32,"end_line":335,"end_character":64},"in_reply_to":"3aaa91ec_85bc0dbc","updated":"2016-06-24 21:10:25.000000000","message":"Assuming this would be done as an L2 agent extension, there\u0027s two cases:\n1) OVS agent detected the port added to br-int, fetches the subport data, and tries to create the vlan interfaces on the tap.\n2) The OVS agent receives an RPC notification that a new subport needs to be created and tries to creat the vlan interface on the tap.\n\nBoth of these cases seem like it\u0027s possible to have a race, so there probably needs to be some kind of retry mechanism. If it just gives up, it will need to reattempt that event later.","commit_id":"94417d5995367aad00912b6d6754d5961134b6a9"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"cf613c5e2bd14341ae33fdc7a115a16afa42979a","unresolved":false,"context_lines":[{"line_number":443,"context_line":"When Nova deletes a VM, it deletes the corresponding Neutron port only if it was"},{"line_number":444,"context_line":"created when booting the VM. In the vlan-aware-vm case the parent port is passed to Nova, so"},{"line_number":445,"context_line":"the port data will remain in the DB after the VM deletion. Nova will delete"},{"line_number":446,"context_line":"the VIF of the VM (in the example tap1) as part of the VM termination. The L2 agent"},{"line_number":447,"context_line":"will detect that deletion and notify the Neutron server that the parent port is down."},{"line_number":448,"context_line":"The OVS agent will clean up the corresponding subports as explained in the next paragraph."},{"line_number":449,"context_line":""}],"source_content_type":"text/x-rst","patch_set":7,"id":"3aaa91ec_c07e0369","line":446,"range":{"start_line":446,"start_character":75,"end_line":446,"end_character":77},"updated":"2016-06-24 20:40:00.000000000","message":"I think this should be OVS to be consistent.","commit_id":"94417d5995367aad00912b6d6754d5961134b6a9"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"cf613c5e2bd14341ae33fdc7a115a16afa42979a","unresolved":false,"context_lines":[{"line_number":447,"context_line":"will detect that deletion and notify the Neutron server that the parent port is down."},{"line_number":448,"context_line":"The OVS agent will clean up the corresponding subports as explained in the next paragraph."},{"line_number":449,"context_line":""},{"line_number":450,"context_line":"The deletion of a trunk that it\u0027s used by a VM is not allowed."},{"line_number":451,"context_line":"The trunk can be deleted (leaving the parent port intact) when the parent port is not"},{"line_number":452,"context_line":"used by any VM. After the trunk is deleted, the parent port can also be deleted."},{"line_number":453,"context_line":""}],"source_content_type":"text/x-rst","patch_set":7,"id":"3aaa91ec_0095ab85","line":450,"range":{"start_line":450,"start_character":29,"end_line":450,"end_character":33},"updated":"2016-06-24 20:40:00.000000000","message":"it\u0027s -\u003e is","commit_id":"94417d5995367aad00912b6d6754d5961134b6a9"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"cf613c5e2bd14341ae33fdc7a115a16afa42979a","unresolved":false,"context_lines":[{"line_number":466,"context_line":""},{"line_number":467,"context_line":"This implementation is based on this `_etherpad \u003chttps://etherpad.openstack.org/p/trunk-bridge-tagged-patch-experiment\u003e`."},{"line_number":468,"context_line":"Credits to Bence Romsics."},{"line_number":469,"context_line":"If use_veth_interconnection\u003dtrue instead of using patch port, veth pairs will be used."},{"line_number":470,"context_line":""},{"line_number":471,"context_line":"::"},{"line_number":472,"context_line":""}],"source_content_type":"text/x-rst","patch_set":7,"id":"3aaa91ec_654921d0","line":469,"range":{"start_line":469,"start_character":0,"end_line":469,"end_character":86},"updated":"2016-06-24 20:40:00.000000000","message":"I think we had somewhat of a consensus in http://lists.openstack.org/pipermail/openstack-dev/2016-June/097025.html that we didn\u0027t need to support this option. Is that right?","commit_id":"94417d5995367aad00912b6d6754d5961134b6a9"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"dd3d1b0a427cc461bd36a7cf5f2744aa89d2ba39","unresolved":false,"context_lines":[{"line_number":466,"context_line":""},{"line_number":467,"context_line":"This implementation is based on this `_etherpad \u003chttps://etherpad.openstack.org/p/trunk-bridge-tagged-patch-experiment\u003e`."},{"line_number":468,"context_line":"Credits to Bence Romsics."},{"line_number":469,"context_line":"If use_veth_interconnection\u003dtrue instead of using patch port, veth pairs will be used."},{"line_number":470,"context_line":""},{"line_number":471,"context_line":"::"},{"line_number":472,"context_line":""}],"source_content_type":"text/x-rst","patch_set":7,"id":"3aaa91ec_811cc86b","line":469,"range":{"start_line":469,"start_character":0,"end_line":469,"end_character":86},"in_reply_to":"3aaa91ec_654921d0","updated":"2016-06-27 13:52:53.000000000","message":"let\u0027s leave a reminder here until the deprecation patch gets merged","commit_id":"94417d5995367aad00912b6d6754d5961134b6a9"},{"author":{"_account_id":7448,"name":"Carl Baldwin","email":"carl@ecbaldwin.net","username":"carl-baldwin"},"change_message_id":"ceb85363e740e3dc3601aeafb3f7fe6faed42939","unresolved":false,"context_lines":[{"line_number":480,"context_line":"         |    tap1                        |"},{"line_number":481,"context_line":"         |          br-trunk-1            |"},{"line_number":482,"context_line":"         |                                |"},{"line_number":483,"context_line":"         | tp-patch-trunk  sp-patch-trunk |"},{"line_number":484,"context_line":"         +-----+-----------------+--------+"},{"line_number":485,"context_line":"               |                 |"},{"line_number":486,"context_line":"               |                 |"}],"source_content_type":"text/x-rst","patch_set":7,"id":"3aaa91ec_e0985f6c","line":483,"range":{"start_line":483,"start_character":27,"end_line":483,"end_character":41},"updated":"2016-06-24 20:54:38.000000000","message":"I assume that there must be something in this subport name on this bridge that distinguishes it from all of the other subports.  Should the name be something like this:\n\n  sp-\u003cvlan id\u003e-patch-trunk","commit_id":"94417d5995367aad00912b6d6754d5961134b6a9"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"72c7a56517044d18b1de70fbcb8edb1efc8a14c3","unresolved":false,"context_lines":[{"line_number":480,"context_line":"         |    tap1                        |"},{"line_number":481,"context_line":"         |          br-trunk-1            |"},{"line_number":482,"context_line":"         |                                |"},{"line_number":483,"context_line":"         | tp-patch-trunk  sp-patch-trunk |"},{"line_number":484,"context_line":"         +-----+-----------------+--------+"},{"line_number":485,"context_line":"               |                 |"},{"line_number":486,"context_line":"               |                 |"}],"source_content_type":"text/x-rst","patch_set":7,"id":"3aaa91ec_7417e744","line":483,"range":{"start_line":483,"start_character":27,"end_line":483,"end_character":41},"in_reply_to":"3aaa91ec_867b747a","updated":"2016-06-27 09:55:19.000000000","message":"it will be distinguished by its own port uuid.","commit_id":"94417d5995367aad00912b6d6754d5961134b6a9"},{"author":{"_account_id":7448,"name":"Carl Baldwin","email":"carl@ecbaldwin.net","username":"carl-baldwin"},"change_message_id":"4627741365d456dca2ae145ed0e1a61f909ab718","unresolved":false,"context_lines":[{"line_number":480,"context_line":"         |    tap1                        |"},{"line_number":481,"context_line":"         |          br-trunk-1            |"},{"line_number":482,"context_line":"         |                                |"},{"line_number":483,"context_line":"         | tp-patch-trunk  sp-patch-trunk |"},{"line_number":484,"context_line":"         +-----+-----------------+--------+"},{"line_number":485,"context_line":"               |                 |"},{"line_number":486,"context_line":"               |                 |"}],"source_content_type":"text/x-rst","patch_set":7,"id":"3aaa91ec_867b747a","line":483,"range":{"start_line":483,"start_character":27,"end_line":483,"end_character":41},"in_reply_to":"3aaa91ec_e0985f6c","updated":"2016-06-24 21:04:57.000000000","message":"To be clear, I don\u0027t think the devref necessarily has to have this level of detail.  I\u0027m just stating the assumptions that I made while reading.","commit_id":"94417d5995367aad00912b6d6754d5961134b6a9"},{"author":{"_account_id":7448,"name":"Carl Baldwin","email":"carl@ecbaldwin.net","username":"carl-baldwin"},"change_message_id":"ceb85363e740e3dc3601aeafb3f7fe6faed42939","unresolved":false,"context_lines":[{"line_number":486,"context_line":"               |                 |"},{"line_number":487,"context_line":"               |                 |"},{"line_number":488,"context_line":"         +-----+-----------------+---------+"},{"line_number":489,"context_line":"         | tp-patch-int     sp-patch-int   |"},{"line_number":490,"context_line":"         |                                 |"},{"line_number":491,"context_line":"         |           br-int                |"},{"line_number":492,"context_line":"         +---------------------------------+"}],"source_content_type":"text/x-rst","patch_set":7,"id":"3aaa91ec_001f6b08","line":489,"range":{"start_line":489,"start_character":11,"end_line":489,"end_character":40},"updated":"2016-06-24 20:54:38.000000000","message":"I assume these names need to be unique for each trunk port and each subport.  So, these names will probably have to be something like:\n\n  tp-\u003ctrunk id\u003e-patch-int\n  sp-\u003ctrunk id\u003e-\u003cvlan id\u003e-patch-int\n\nOr, something like that.","commit_id":"94417d5995367aad00912b6d6754d5961134b6a9"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"72c7a56517044d18b1de70fbcb8edb1efc8a14c3","unresolved":false,"context_lines":[{"line_number":486,"context_line":"               |                 |"},{"line_number":487,"context_line":"               |                 |"},{"line_number":488,"context_line":"         +-----+-----------------+---------+"},{"line_number":489,"context_line":"         | tp-patch-int     sp-patch-int   |"},{"line_number":490,"context_line":"         |                                 |"},{"line_number":491,"context_line":"         |           br-int                |"},{"line_number":492,"context_line":"         +---------------------------------+"}],"source_content_type":"text/x-rst","patch_set":7,"id":"3aaa91ec_343f1fb5","line":489,"range":{"start_line":489,"start_character":11,"end_line":489,"end_character":40},"in_reply_to":"3aaa91ec_001f6b08","updated":"2016-06-27 09:55:19.000000000","message":"yes this was discusded in a previous revision.\nsee my first comment on line 457 on patch set 2:\nhttps://review.openstack.org/#/c/318317/2/doc/source/devref/openvswitch_agent.rst@457\n\nthe intention is to use part of the port uuid to make the names unique. for clarity of the diagram it is ommited.","commit_id":"94417d5995367aad00912b6d6754d5961134b6a9"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"cf613c5e2bd14341ae33fdc7a115a16afa42979a","unresolved":false,"context_lines":[{"line_number":503,"context_line":"A VM is spawned passing to Nova the port-id of a parent port associated with"},{"line_number":504,"context_line":"a trunk. Neutron will pass to Nova the bridge where to plug the vif as part of the vif details."},{"line_number":505,"context_line":"The os-vif driver creates the trunk bridge br-trunk-1 if it does not exist in plug()"},{"line_number":506,"context_line":"with the parent port ID in the external-ids field. It will create the tap interface"},{"line_number":507,"context_line":"tap1 and plug it into br-trunk-1 setting the parent port ID in the external-ids."},{"line_number":508,"context_line":"The OVS agent will be monitoring the creation of ports on the trunk bridges. When it detects"},{"line_number":509,"context_line":"that a new port has been created on the trunk bridge, it will do the following:"}],"source_content_type":"text/x-rst","patch_set":7,"id":"3aaa91ec_a5666927","line":506,"range":{"start_line":506,"start_character":0,"end_line":506,"end_character":49},"updated":"2016-06-24 20:40:00.000000000","message":"I think because the OVS agent will be responsible for deleting the trunk bridge, the OVS agent should be setting the parent port ID in the bridge\u0027s external_ids field. Nova/os-vif could do this, but it would add extra complexity into os-vif because it then has to check if the bridge already exists before it creates it and determine if it needs to set the external_ids based upon its preexistence. Then the OVS agent would have to rely upon that assumed convention rather than handle it itself.\n\nWe should remove this snippet.","commit_id":"94417d5995367aad00912b6d6754d5961134b6a9"},{"author":{"_account_id":7448,"name":"Carl Baldwin","email":"carl@ecbaldwin.net","username":"carl-baldwin"},"change_message_id":"ceb85363e740e3dc3601aeafb3f7fe6faed42939","unresolved":false,"context_lines":[{"line_number":503,"context_line":"A VM is spawned passing to Nova the port-id of a parent port associated with"},{"line_number":504,"context_line":"a trunk. Neutron will pass to Nova the bridge where to plug the vif as part of the vif details."},{"line_number":505,"context_line":"The os-vif driver creates the trunk bridge br-trunk-1 if it does not exist in plug()"},{"line_number":506,"context_line":"with the parent port ID in the external-ids field. It will create the tap interface"},{"line_number":507,"context_line":"tap1 and plug it into br-trunk-1 setting the parent port ID in the external-ids."},{"line_number":508,"context_line":"The OVS agent will be monitoring the creation of ports on the trunk bridges. When it detects"},{"line_number":509,"context_line":"that a new port has been created on the trunk bridge, it will do the following:"}],"source_content_type":"text/x-rst","patch_set":7,"id":"3aaa91ec_80d6bb4c","line":506,"range":{"start_line":506,"start_character":0,"end_line":506,"end_character":49},"updated":"2016-06-24 20:54:38.000000000","message":"This doesn\u0027t match the patch that Rawlin has up [1].  I prefer what\u0027s in that patch.  The os-vif driver will not add the parent port ID in the external-ids field.  The ovs agent will get the parent port from the port instead of the bridge.\n\n [1] https://review.openstack.org/#/c/330818","commit_id":"94417d5995367aad00912b6d6754d5961134b6a9"},{"author":{"_account_id":7448,"name":"Carl Baldwin","email":"carl@ecbaldwin.net","username":"carl-baldwin"},"change_message_id":"4627741365d456dca2ae145ed0e1a61f909ab718","unresolved":false,"context_lines":[{"line_number":503,"context_line":"A VM is spawned passing to Nova the port-id of a parent port associated with"},{"line_number":504,"context_line":"a trunk. Neutron will pass to Nova the bridge where to plug the vif as part of the vif details."},{"line_number":505,"context_line":"The os-vif driver creates the trunk bridge br-trunk-1 if it does not exist in plug()"},{"line_number":506,"context_line":"with the parent port ID in the external-ids field. It will create the tap interface"},{"line_number":507,"context_line":"tap1 and plug it into br-trunk-1 setting the parent port ID in the external-ids."},{"line_number":508,"context_line":"The OVS agent will be monitoring the creation of ports on the trunk bridges. When it detects"},{"line_number":509,"context_line":"that a new port has been created on the trunk bridge, it will do the following:"}],"source_content_type":"text/x-rst","patch_set":7,"id":"3aaa91ec_c6c73c94","line":506,"range":{"start_line":506,"start_character":0,"end_line":506,"end_character":49},"in_reply_to":"3aaa91ec_80d6bb4c","updated":"2016-06-24 21:04:57.000000000","message":"BTW, this comment is the reason for my -1.","commit_id":"94417d5995367aad00912b6d6754d5961134b6a9"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"72c7a56517044d18b1de70fbcb8edb1efc8a14c3","unresolved":false,"context_lines":[{"line_number":503,"context_line":"A VM is spawned passing to Nova the port-id of a parent port associated with"},{"line_number":504,"context_line":"a trunk. Neutron will pass to Nova the bridge where to plug the vif as part of the vif details."},{"line_number":505,"context_line":"The os-vif driver creates the trunk bridge br-trunk-1 if it does not exist in plug()"},{"line_number":506,"context_line":"with the parent port ID in the external-ids field. It will create the tap interface"},{"line_number":507,"context_line":"tap1 and plug it into br-trunk-1 setting the parent port ID in the external-ids."},{"line_number":508,"context_line":"The OVS agent will be monitoring the creation of ports on the trunk bridges. When it detects"},{"line_number":509,"context_line":"that a new port has been created on the trunk bridge, it will do the following:"}],"source_content_type":"text/x-rst","patch_set":7,"id":"3aaa91ec_518c7dd9","line":506,"range":{"start_line":506,"start_character":0,"end_line":506,"end_character":49},"in_reply_to":"3aaa91ec_c6c73c94","updated":"2016-06-27 09:55:19.000000000","message":"the original intention was that it would be on both so that when the port was deleted the ovs agent could use the port id to identify what bridge to delete.\n\nwith the current patch its up to the agent to determine how to track the correlation between trunk port and then trunk bridge on deletion of the trunk port.","commit_id":"94417d5995367aad00912b6d6754d5961134b6a9"},{"author":{"_account_id":7448,"name":"Carl Baldwin","email":"carl@ecbaldwin.net","username":"carl-baldwin"},"change_message_id":"ceb85363e740e3dc3601aeafb3f7fe6faed42939","unresolved":false,"context_lines":[{"line_number":504,"context_line":"a trunk. Neutron will pass to Nova the bridge where to plug the vif as part of the vif details."},{"line_number":505,"context_line":"The os-vif driver creates the trunk bridge br-trunk-1 if it does not exist in plug()"},{"line_number":506,"context_line":"with the parent port ID in the external-ids field. It will create the tap interface"},{"line_number":507,"context_line":"tap1 and plug it into br-trunk-1 setting the parent port ID in the external-ids."},{"line_number":508,"context_line":"The OVS agent will be monitoring the creation of ports on the trunk bridges. When it detects"},{"line_number":509,"context_line":"that a new port has been created on the trunk bridge, it will do the following:"},{"line_number":510,"context_line":""}],"source_content_type":"text/x-rst","patch_set":7,"id":"3aaa91ec_e0c97fad","line":507,"range":{"start_line":507,"start_character":33,"end_line":507,"end_character":79},"updated":"2016-06-24 20:54:38.000000000","message":"However, this is still true as far as I know.","commit_id":"94417d5995367aad00912b6d6754d5961134b6a9"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"72c7a56517044d18b1de70fbcb8edb1efc8a14c3","unresolved":false,"context_lines":[{"line_number":504,"context_line":"a trunk. Neutron will pass to Nova the bridge where to plug the vif as part of the vif details."},{"line_number":505,"context_line":"The os-vif driver creates the trunk bridge br-trunk-1 if it does not exist in plug()"},{"line_number":506,"context_line":"with the parent port ID in the external-ids field. It will create the tap interface"},{"line_number":507,"context_line":"tap1 and plug it into br-trunk-1 setting the parent port ID in the external-ids."},{"line_number":508,"context_line":"The OVS agent will be monitoring the creation of ports on the trunk bridges. When it detects"},{"line_number":509,"context_line":"that a new port has been created on the trunk bridge, it will do the following:"},{"line_number":510,"context_line":""}],"source_content_type":"text/x-rst","patch_set":7,"id":"3aaa91ec_517abdbd","line":507,"range":{"start_line":507,"start_character":33,"end_line":507,"end_character":79},"in_reply_to":"3aaa91ec_c6573c56","updated":"2016-06-27 09:55:19.000000000","message":"its also done by os-vif when it plugs the port instead of libvirt.","commit_id":"94417d5995367aad00912b6d6754d5961134b6a9"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"34939891858e9076c9af92259f59fc2b61c7bc45","unresolved":false,"context_lines":[{"line_number":504,"context_line":"a trunk. Neutron will pass to Nova the bridge where to plug the vif as part of the vif details."},{"line_number":505,"context_line":"The os-vif driver creates the trunk bridge br-trunk-1 if it does not exist in plug()"},{"line_number":506,"context_line":"with the parent port ID in the external-ids field. It will create the tap interface"},{"line_number":507,"context_line":"tap1 and plug it into br-trunk-1 setting the parent port ID in the external-ids."},{"line_number":508,"context_line":"The OVS agent will be monitoring the creation of ports on the trunk bridges. When it detects"},{"line_number":509,"context_line":"that a new port has been created on the trunk bridge, it will do the following:"},{"line_number":510,"context_line":""}],"source_content_type":"text/x-rst","patch_set":7,"id":"3aaa91ec_c6573c56","line":507,"range":{"start_line":507,"start_character":33,"end_line":507,"end_character":79},"in_reply_to":"3aaa91ec_e0c97fad","updated":"2016-06-24 21:10:25.000000000","message":"Yes, libvirt does this automatically when creating ports on OVS bridges.","commit_id":"94417d5995367aad00912b6d6754d5961134b6a9"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"cf613c5e2bd14341ae33fdc7a115a16afa42979a","unresolved":false,"context_lines":[{"line_number":567,"context_line":"Trunk deletion"},{"line_number":568,"context_line":"++++++++++++++"},{"line_number":569,"context_line":""},{"line_number":570,"context_line":"When Nova deletes a VM, it doesn\u0027t delete the corresponding Neutron port only if it was"},{"line_number":571,"context_line":"created when booting the VM. In the vlan-aware-vm case the parent port is passed to Nova, so"},{"line_number":572,"context_line":"the port data will remain in the DB after the VM deletion. Nova will delete"},{"line_number":573,"context_line":"the port on the trunk bridge where the VM is plugged. The L2 agent"},{"line_number":574,"context_line":"will detect that and delete the trunk bridge. It will notify the Neutron server that the parent"}],"source_content_type":"text/x-rst","patch_set":7,"id":"3aaa91ec_20454745","line":571,"range":{"start_line":570,"start_character":0,"end_line":571,"end_character":27},"updated":"2016-06-24 20:40:00.000000000","message":"\"When Nova deletes a VM, it deletes the VM\u0027s corresponding Neutron ports only if they were created by Nova when booting the VM.\"\n\nDitto L443.","commit_id":"94417d5995367aad00912b6d6754d5961134b6a9"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"dd3d1b0a427cc461bd36a7cf5f2744aa89d2ba39","unresolved":false,"context_lines":[{"line_number":567,"context_line":"Trunk deletion"},{"line_number":568,"context_line":"++++++++++++++"},{"line_number":569,"context_line":""},{"line_number":570,"context_line":"When Nova deletes a VM, it doesn\u0027t delete the corresponding Neutron port only if it was"},{"line_number":571,"context_line":"created when booting the VM. In the vlan-aware-vm case the parent port is passed to Nova, so"},{"line_number":572,"context_line":"the port data will remain in the DB after the VM deletion. Nova will delete"},{"line_number":573,"context_line":"the port on the trunk bridge where the VM is plugged. The L2 agent"},{"line_number":574,"context_line":"will detect that and delete the trunk bridge. It will notify the Neutron server that the parent"}],"source_content_type":"text/x-rst","patch_set":7,"id":"3aaa91ec_e1382c40","line":571,"range":{"start_line":570,"start_character":0,"end_line":571,"end_character":27},"in_reply_to":"3aaa91ec_20454745","updated":"2016-06-27 13:52:53.000000000","message":"Done","commit_id":"94417d5995367aad00912b6d6754d5961134b6a9"},{"author":{"_account_id":16800,"name":"Rawlin Peters","email":"rawlin.peters@hpe.com","username":"rawlin"},"change_message_id":"cf613c5e2bd14341ae33fdc7a115a16afa42979a","unresolved":false,"context_lines":[{"line_number":574,"context_line":"will detect that and delete the trunk bridge. It will notify the Neutron server that the parent"},{"line_number":575,"context_line":"port is down."},{"line_number":576,"context_line":""},{"line_number":577,"context_line":"The deletion of a trunk that it\u0027s used by a VM is not allowed."},{"line_number":578,"context_line":"The trunk can be deleted (leaving the parent port intact) when the parent port is not"},{"line_number":579,"context_line":"used by any VM. After the trunk is deleted, the parent port can also be deleted."},{"line_number":580,"context_line":""}],"source_content_type":"text/x-rst","patch_set":7,"id":"3aaa91ec_a077779e","line":577,"range":{"start_line":577,"start_character":29,"end_line":577,"end_character":33},"updated":"2016-06-24 20:40:00.000000000","message":"it\u0027s -\u003e is","commit_id":"94417d5995367aad00912b6d6754d5961134b6a9"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"3baece6d1f96cbf9ef3898d685ad1a92818e81a8","unresolved":false,"context_lines":[{"line_number":466,"context_line":""},{"line_number":467,"context_line":"This implementation is based on this `_etherpad \u003chttps://etherpad.openstack.org/p/trunk-bridge-tagged-patch-experiment\u003e`."},{"line_number":468,"context_line":"Credits to Bence Romsics."},{"line_number":469,"context_line":"The option use_veth_interconnection\u003dtrue won\u0027t be supported, it will probably be deprecated soon,"},{"line_number":470,"context_line":"see [1]."},{"line_number":471,"context_line":""},{"line_number":472,"context_line":"::"}],"source_content_type":"text/x-rst","patch_set":8,"id":"3aaa91ec_4b53cf39","line":469,"updated":"2016-06-28 17:49:02.000000000","message":"assumed this merge as is, this should probably read as follows:\n\nThe option use_veth_interconnection\u003dtrue won\u0027t be supported as it may be deprecated in the future. For an update on the deprecation process see [1].\n\nBut it\u0027s a nit.","commit_id":"8817c7d14daea6e36d5f1a8ca5c35f537d3a900b"}]}
