)]}'
{"doc/source/install/advanced_deployment/distributed_compute_node.rst":[{"author":{"_account_id":6926,"name":"Bogdan Dobrelya","email":"bdobreli@redhat.com","username":"bogdando"},"change_message_id":"c2fe7825c53c1e002345cb35e0e4b798bbbfdea1","unresolved":false,"context_lines":[{"line_number":97,"context_line":"control plane."},{"line_number":98,"context_line":""},{"line_number":99,"context_line":".. note:: there is no automation provided for such DCN network setups,"},{"line_number":100,"context_line":"   so you have to have those relays and/or edge DHCP agents configured manually."},{"line_number":101,"context_line":"   Also, using OVN might simplify setups as it does not require running"},{"line_number":102,"context_line":"   Neutron DHCP/L3 agents at all."},{"line_number":103,"context_line":""}],"source_content_type":"text/x-rst","patch_set":1,"id":"3fce034c_7f3bb7d0","line":100,"range":{"start_line":100,"start_character":43,"end_line":100,"end_character":59},"updated":"2019-04-12 13:23:40.000000000","message":"not sure if that\u0027s correct, we prolly may assign the agent roles to DCN computes w/o any problems","commit_id":"d2ca2c9d7d60220699721407f649be0de6b1917f"},{"author":{"_account_id":12398,"name":"Dan Sneddon","email":"dsneddon@redhat.com","username":"dsneddon"},"change_message_id":"395e9f2c12025e5b3bc8c8d0219e484dbf042b34","unresolved":false,"context_lines":[{"line_number":50,"context_line":"Loss of an edge site"},{"line_number":51,"context_line":"^^^^^^^^^^^^^^^^^^^^"},{"line_number":52,"context_line":""},{"line_number":53,"context_line":"Running Nova VM instances will keep running. If stopped running, you need the"},{"line_number":54,"context_line":"control plane back to recover the stopped or crashed workloads. If Neutron DHCP"},{"line_number":55,"context_line":"agent is centralized, and we are forwarding DHCP requests to the central site,"},{"line_number":56,"context_line":"any VMs that are trying to renew their IPs will eventually time out and lose"}],"source_content_type":"text/x-rst","patch_set":2,"id":"3fce034c_594cad9d","line":53,"range":{"start_line":53,"start_character":45,"end_line":53,"end_character":63},"updated":"2019-04-15 22:38:18.000000000","message":"s/If stopped running/If the instances stop running for any reason/","commit_id":"58d5da3f721503f54dde683c23419112ab0c7bfb"},{"author":{"_account_id":12398,"name":"Dan Sneddon","email":"dsneddon@redhat.com","username":"dsneddon"},"change_message_id":"395e9f2c12025e5b3bc8c8d0219e484dbf042b34","unresolved":false,"context_lines":[{"line_number":116,"context_line":""},{"line_number":117,"context_line":"The simplest solution we recommend for DCN would involve only external provider"},{"line_number":118,"context_line":"networks at the edge. For that case, it is also recommended to use config-drive"},{"line_number":119,"context_line":"or another configuration mechanism other than cloud-init. Otherwise, the latter"},{"line_number":120,"context_line":"requires a `169.254.169.254/32` route for the provider routers to forward data"},{"line_number":121,"context_line":"to the metadata service."},{"line_number":122,"context_line":""}],"source_content_type":"text/x-rst","patch_set":2,"id":"3fce034c_d9765dd9","line":119,"range":{"start_line":119,"start_character":46,"end_line":119,"end_character":56},"updated":"2019-04-15 22:38:18.000000000","message":"This might be misleading. IIUC a VM image that is configured for cloud-init will mount and read the config-drive image. So cloud-init is involved, but the metadata IP of 169.254.169.254 is not used.\n\nOr do I have a misunderstanding about how config-drive would be mounted?","commit_id":"58d5da3f721503f54dde683c23419112ab0c7bfb"},{"author":{"_account_id":6926,"name":"Bogdan Dobrelya","email":"bdobreli@redhat.com","username":"bogdando"},"change_message_id":"2db34ee6958ce87e098b9a92646d6bcddbd6e4cb","unresolved":false,"context_lines":[{"line_number":116,"context_line":""},{"line_number":117,"context_line":"The simplest solution we recommend for DCN would involve only external provider"},{"line_number":118,"context_line":"networks at the edge. For that case, it is also recommended to use config-drive"},{"line_number":119,"context_line":"or another configuration mechanism other than cloud-init. Otherwise, the latter"},{"line_number":120,"context_line":"requires a `169.254.169.254/32` route for the provider routers to forward data"},{"line_number":121,"context_line":"to the metadata service."},{"line_number":122,"context_line":""}],"source_content_type":"text/x-rst","patch_set":2,"id":"3fce034c_a146937b","line":119,"range":{"start_line":119,"start_character":46,"end_line":119,"end_character":56},"in_reply_to":"3fce034c_d9765dd9","updated":"2019-04-17 13:33:39.000000000","message":"Done","commit_id":"58d5da3f721503f54dde683c23419112ab0c7bfb"},{"author":{"_account_id":6926,"name":"Bogdan Dobrelya","email":"bdobreli@redhat.com","username":"bogdando"},"change_message_id":"024a9c9cd6571121c7d25e00775c67437994c888","unresolved":false,"context_lines":[{"line_number":95,"context_line":"routing) on its own. While for traditional or"},{"line_number":96,"context_line":"`Routed Provider Networks \u003chttps://docs.openstack.org/neutron/latest/admin/config-routed-networks.html\u003e`_,"},{"line_number":97,"context_line":"where DHCP agents are placed on controllers at the central site, that may be"},{"line_number":98,"context_line":"either a DHCP relay on every provider network. Or DHCP agents need to be"},{"line_number":99,"context_line":"scheduled at the edge sites, when there is no L2 connectivity to the central"},{"line_number":100,"context_line":"control plane. Such setups also require highly reliable links between remote and"},{"line_number":101,"context_line":"central sites."},{"line_number":102,"context_line":""},{"line_number":103,"context_line":".. note::  Neither of DHCP relays/agents at compute nodes, nor routed/external"}],"source_content_type":"text/x-rst","patch_set":4,"id":"3fce034c_77f13095","line":100,"range":{"start_line":98,"start_character":7,"end_line":100,"end_character":13},"updated":"2019-04-18 15:14:06.000000000","message":"this may be not super accurate, citing Dan Smith: \"depends on where your relay is, but the point of dhcp relay is to forward it over an L3 link instead of requiring L2 across the wan\"\n\nbut perhaps we could accept that change and polish all that in follow-ups...","commit_id":"f856216f9d2e2472e8fdff774539bd13756c1473"},{"author":{"_account_id":6926,"name":"Bogdan Dobrelya","email":"bdobreli@redhat.com","username":"bogdando"},"change_message_id":"0bdb0d62a620a120058d907d63d14c3cfc510c0d","unresolved":false,"context_lines":[{"line_number":95,"context_line":"routing) on its own. While for traditional or"},{"line_number":96,"context_line":"`Routed Provider Networks \u003chttps://docs.openstack.org/neutron/latest/admin/config-routed-networks.html\u003e`_,"},{"line_number":97,"context_line":"where DHCP agents are placed on controllers at the central site, that may be"},{"line_number":98,"context_line":"either a DHCP relay on every provider network. Or DHCP agents need to be"},{"line_number":99,"context_line":"scheduled at the edge sites, when there is no L2 connectivity to the central"},{"line_number":100,"context_line":"control plane. Such setups also require highly reliable links between remote and"},{"line_number":101,"context_line":"central sites."},{"line_number":102,"context_line":""},{"line_number":103,"context_line":".. note::  Neither of DHCP relays/agents at compute nodes, nor routed/external"}],"source_content_type":"text/x-rst","patch_set":4,"id":"3fce034c_d44570b5","line":100,"range":{"start_line":98,"start_character":7,"end_line":100,"end_character":13},"in_reply_to":"3fce034c_387394b3","updated":"2019-04-19 08:03:37.000000000","message":"Done","commit_id":"f856216f9d2e2472e8fdff774539bd13756c1473"},{"author":{"_account_id":12398,"name":"Dan Sneddon","email":"dsneddon@redhat.com","username":"dsneddon"},"change_message_id":"f23bbb644b273e650bfc7ae6202f7bd31cbc88dc","unresolved":false,"context_lines":[{"line_number":95,"context_line":"routing) on its own. While for traditional or"},{"line_number":96,"context_line":"`Routed Provider Networks \u003chttps://docs.openstack.org/neutron/latest/admin/config-routed-networks.html\u003e`_,"},{"line_number":97,"context_line":"where DHCP agents are placed on controllers at the central site, that may be"},{"line_number":98,"context_line":"either a DHCP relay on every provider network. Or DHCP agents need to be"},{"line_number":99,"context_line":"scheduled at the edge sites, when there is no L2 connectivity to the central"},{"line_number":100,"context_line":"control plane. Such setups also require highly reliable links between remote and"},{"line_number":101,"context_line":"central sites."},{"line_number":102,"context_line":""},{"line_number":103,"context_line":".. note::  Neither of DHCP relays/agents at compute nodes, nor routed/external"}],"source_content_type":"text/x-rst","patch_set":4,"id":"3fce034c_387394b3","line":100,"range":{"start_line":98,"start_character":7,"end_line":100,"end_character":13},"in_reply_to":"3fce034c_77f13095","updated":"2019-04-18 19:45:22.000000000","message":"Perhaps we can rephrase. DHCP relay is inherently layer 3, but you still need a DHCP relay to be set on the router for every provider network, or you need local DHCP servers so that no relay servers are needed.\n\nIf you have layer 2 connectivity back to the central site, then you don\u0027t need DHCP relay if you are not using Neutron routed networks.","commit_id":"f856216f9d2e2472e8fdff774539bd13756c1473"}]}