)]}'
{"/COMMIT_MSG":[{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"bf01f5a32b030255385c7bdafbb3f01e899df7aa","unresolved":false,"context_lines":[{"line_number":8,"context_line":""},{"line_number":9,"context_line":"The VIFVHostUser class does not provide all the necessary information"},{"line_number":10,"context_line":"to perform plug/unplug for fast path vhostuser ports."},{"line_number":11,"context_line":"This patch defines a new class for this specific case that adds the"},{"line_number":12,"context_line":"necessary fields."},{"line_number":13,"context_line":""},{"line_number":14,"context_line":"Change-Id: Ibae8e2ae145b2b9ad2b4955825638dee9b56f0aa"},{"line_number":15,"context_line":"Signed-off-by: Francesco Santoro \u003cfrancesco.santoro@6wind.com\u003e"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"da6895a0_8c36c001","line":12,"range":{"start_line":11,"start_character":0,"end_line":12,"end_character":17},"updated":"2016-10-22 14:03:26.000000000","message":"looking at the patch i would prefer to see the new field added to the current VIFVHostUser class as it will allow\nus to skip generating the name here\nhttps://github.com/openstack/os-vif/blob/ca5d7d6ff170f734429d3fbebd29dbe3fa5880da/vif_plug_ovs/ovs.py#L102\nand pass it from nova/neutron in the feuture","commit_id":"bc0dd254c5f6221ce10cf59efd4b260139eac4ed"},{"author":{"_account_id":17671,"name":"Francesco Santoro","email":"fraroot@gmail.com","username":"francesco.santoro"},"change_message_id":"e4996ad648e006c2af6ca4ad6b2cf9101831fd85","unresolved":false,"context_lines":[{"line_number":8,"context_line":""},{"line_number":9,"context_line":"The VIFVHostUser class does not provide all the necessary information"},{"line_number":10,"context_line":"to perform plug/unplug for fast path vhostuser ports."},{"line_number":11,"context_line":"This patch defines a new class for this specific case that adds the"},{"line_number":12,"context_line":"necessary fields."},{"line_number":13,"context_line":""},{"line_number":14,"context_line":"Change-Id: Ibae8e2ae145b2b9ad2b4955825638dee9b56f0aa"},{"line_number":15,"context_line":"Signed-off-by: Francesco Santoro \u003cfrancesco.santoro@6wind.com\u003e"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"da6895a0_9b1c2b15","line":12,"range":{"start_line":11,"start_character":0,"end_line":12,"end_character":17},"in_reply_to":"da6895a0_8c36c001","updated":"2016-10-24 07:46:34.000000000","message":"No problem. If the vif_name is a needed field I will add it to the base class","commit_id":"bc0dd254c5f6221ce10cf59efd4b260139eac4ed"}],"doc/source/vif_types.rst":[{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"3722f5c13540e0a23e8378ca66da5c4a8bd51b5e","unresolved":false,"context_lines":[{"line_number":61,"context_line":"virtio-net vhost protocol on this socket in whatever means is most"},{"line_number":62,"context_line":"suitable."},{"line_number":63,"context_line":""},{"line_number":64,"context_line":"VIFVHostUserFP"},{"line_number":65,"context_line":"--------------"},{"line_number":66,"context_line":""},{"line_number":67,"context_line":"This class provides the same libvirt configuration used for the VIFVHostUser"},{"line_number":68,"context_line":"object when the fast path vhostuser backend is used."},{"line_number":69,"context_line":"In particular, this class is needed to allow differentiation for"},{"line_number":70,"context_line":"fast path vhostuser (that could need additional specific fields)"},{"line_number":71,"context_line":"compared to the default vhostuser ovs case."},{"line_number":72,"context_line":""},{"line_number":73,"context_line":"VIFHostDevice"},{"line_number":74,"context_line":"-------------"}],"source_content_type":"text/x-rst","patch_set":11,"id":"1a6eadb0_6ecfc0b0","line":71,"range":{"start_line":64,"start_character":0,"end_line":71,"end_character":43},"updated":"2016-12-13 01:29:18.000000000","message":"this is not required and should not be added.\n\nyou will never need an addtional value in the vif type itself.\n\nvhost user as a protocal only requires two valuse for the virtual interface the unix socket and the mode.\n\ntechnincally its 3 socket domain, socket path/uri and mode but since unix sockets are teh only supporte socket domain its two. if that ever changes the base VIFVHostUser class will be extended.","commit_id":"8847c488d057db67a158150313f01dadfc05fba6"}],"os_vif/objects/vif.py":[{"author":{"_account_id":1779,"name":"Daniel Berrange","email":"berrange@redhat.com","username":"berrange"},"change_message_id":"9a1ec932c1dc44db50dc65531f57b2205d302fb6","unresolved":false,"context_lines":[{"line_number":134,"context_line":"        \u0027vif_name\u0027: fields.StringField(),"},{"line_number":135,"context_line":""},{"line_number":136,"context_line":"        # Type of plug fast path should perform"},{"line_number":137,"context_line":"        \u0027plug_type\u0027: fields.StringField(),"},{"line_number":138,"context_line":""},{"line_number":139,"context_line":"        # Name of the bridge to connect to"},{"line_number":140,"context_line":"        \u0027bridge_name\u0027: fields.StringField(),"}],"source_content_type":"text/x-python","patch_set":1,"id":"5a8bc5a2_7cc81923","line":137,"updated":"2016-09-23 14:54:17.000000000","message":"Based on the description, this sounds like a enum. Can you explain this more. Having a type field that implies 2 different behaviours tends to suggest the need for multiple classes.","commit_id":"0d237c70b5fd444f8b446854b35c61028c04f6e4"},{"author":{"_account_id":1779,"name":"Daniel Berrange","email":"berrange@redhat.com","username":"berrange"},"change_message_id":"d1f46f94c8d288018540a817b479184d12352221","unresolved":false,"context_lines":[{"line_number":134,"context_line":"        \u0027vif_name\u0027: fields.StringField(),"},{"line_number":135,"context_line":""},{"line_number":136,"context_line":"        # Type of plug fast path should perform"},{"line_number":137,"context_line":"        \u0027plug_type\u0027: fields.StringField(),"},{"line_number":138,"context_line":""},{"line_number":139,"context_line":"        # Name of the bridge to connect to"},{"line_number":140,"context_line":"        \u0027bridge_name\u0027: fields.StringField(),"}],"source_content_type":"text/x-python","patch_set":1,"id":"5a8bc5a2_42f45485","line":137,"in_reply_to":"5a8bc5a2_02988cbf","updated":"2016-09-23 15:22:23.000000000","message":"That\u0027s sounding rather alot more complicated that what is currently supported in Nova. From a modelling pov we don\u0027t want to have to add the union of attributes for all those vif types into the same object.\n\nI think I\u0027d like to see more background design info on FP vhostuser before we starting adding arbitrary new objects here. Where is this code you\u0027ve reworked for nova. I don\u0027t see any open patches from you in gerrit related to nova.","commit_id":"0d237c70b5fd444f8b446854b35c61028c04f6e4"},{"author":{"_account_id":17671,"name":"Francesco Santoro","email":"fraroot@gmail.com","username":"francesco.santoro"},"change_message_id":"ccdca5ed35c247c7f5d96912ccfbf9da9f0cdae6","unresolved":false,"context_lines":[{"line_number":134,"context_line":"        \u0027vif_name\u0027: fields.StringField(),"},{"line_number":135,"context_line":""},{"line_number":136,"context_line":"        # Type of plug fast path should perform"},{"line_number":137,"context_line":"        \u0027plug_type\u0027: fields.StringField(),"},{"line_number":138,"context_line":""},{"line_number":139,"context_line":"        # Name of the bridge to connect to"},{"line_number":140,"context_line":"        \u0027bridge_name\u0027: fields.StringField(),"}],"source_content_type":"text/x-python","patch_set":1,"id":"5a8bc5a2_c86fa3c3","line":137,"in_reply_to":"5a8bc5a2_42f45485","updated":"2016-09-23 15:48:38.000000000","message":"For testing purposes I am using the networking_6wind plugin and a monkey patch for nova. Refer to these patches: https://review.openstack.org/#/c/375577/ and https://review.openstack.org/#/c/366108/4.\nfast path has similar needs than ovs-dpdk case where the vhostuser vif type is not enough to perform the plug since some ovs info are needed.","commit_id":"0d237c70b5fd444f8b446854b35c61028c04f6e4"},{"author":{"_account_id":17671,"name":"Francesco Santoro","email":"fraroot@gmail.com","username":"francesco.santoro"},"change_message_id":"30d2fdb10e2c633b61bc3afaccbf7cd27e2d8f30","unresolved":false,"context_lines":[{"line_number":134,"context_line":"        \u0027vif_name\u0027: fields.StringField(),"},{"line_number":135,"context_line":""},{"line_number":136,"context_line":"        # Type of plug fast path should perform"},{"line_number":137,"context_line":"        \u0027plug_type\u0027: fields.StringField(),"},{"line_number":138,"context_line":""},{"line_number":139,"context_line":"        # Name of the bridge to connect to"},{"line_number":140,"context_line":"        \u0027bridge_name\u0027: fields.StringField(),"}],"source_content_type":"text/x-python","patch_set":1,"id":"5a8bc5a2_02988cbf","line":137,"in_reply_to":"5a8bc5a2_7cc81923","updated":"2016-09-23 15:12:15.000000000","message":"Yes, it could be an enum. Actually fast path vhostuser ports can be plugged in ovs, linuxbrige, calico, midonet. The current implementation in nova (plug_vhostuser_fp) only performs ovs plug but the os-vif driver should be able to perform all the mentioned scenarios. I have patches for nova that reworks code and add this \"fp_plug_type\" in the vif details to allow our mechanism drivers choose which kind of plug should be used.","commit_id":"0d237c70b5fd444f8b446854b35c61028c04f6e4"},{"author":{"_account_id":17671,"name":"Francesco Santoro","email":"fraroot@gmail.com","username":"francesco.santoro"},"change_message_id":"db64466ff61a7860e8962573d8b7458368b27cfb","unresolved":false,"context_lines":[{"line_number":134,"context_line":"        \u0027vif_name\u0027: fields.StringField(),"},{"line_number":135,"context_line":""},{"line_number":136,"context_line":"        # Type of plug fast path should perform"},{"line_number":137,"context_line":"        \u0027plug_type\u0027: fields.StringField(),"},{"line_number":138,"context_line":""},{"line_number":139,"context_line":"        # Name of the bridge to connect to"},{"line_number":140,"context_line":"        \u0027bridge_name\u0027: fields.StringField(),"}],"source_content_type":"text/x-python","patch_set":1,"id":"5a8bc5a2_f2d5ad42","line":137,"in_reply_to":"5a8bc5a2_93d6474d","updated":"2016-09-27 12:05:53.000000000","message":"I just wanted to propose the less intrusive change trying to delegate the dirty work within the plugin (checking the corresponding fields according to the plug_type value, authorizing leaving some of them unset).\nI have to admit that a \u0027plug_type\u0027 field can be misunderstood and generate confusion.\n\nAs you said some of the fields are meaningful just in some scenarios, thus having multiple classes will be the cleanest approach but, if my understanding is correct, Daniel is not so happy with it from a design perspective (mixing of vif types).\nI am totally open to an alternative approach that fit the design logic and rationale of os-vif.\n\nThat said the current implementation of os-vif makes quite hard the migration of nova plug_vhostuser_fp to an os-vif driver. In particular when hybrid_plug is active since the hybrid bridge name information is not exported in the VIFVHostUser object.\nMaybe this patch is adding \"too much\", but even the basic scenario with ovs + security groups + fastpath vhostuser won\u0027t work (unless you have some ideas I am missing).","commit_id":"0d237c70b5fd444f8b446854b35c61028c04f6e4"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"ed269c2b430150ee3282d16d0c9f35b30ce4270a","unresolved":false,"context_lines":[{"line_number":134,"context_line":"        \u0027vif_name\u0027: fields.StringField(),"},{"line_number":135,"context_line":""},{"line_number":136,"context_line":"        # Type of plug fast path should perform"},{"line_number":137,"context_line":"        \u0027plug_type\u0027: fields.StringField(),"},{"line_number":138,"context_line":""},{"line_number":139,"context_line":"        # Name of the bridge to connect to"},{"line_number":140,"context_line":"        \u0027bridge_name\u0027: fields.StringField(),"}],"source_content_type":"text/x-python","patch_set":1,"id":"5a8bc5a2_93d6474d","line":137,"in_reply_to":"5a8bc5a2_c86fa3c3","updated":"2016-09-27 01:07:16.000000000","message":"i tend to lean towords seperate classes in this case also instead of an enum that said from your discription i think you would need 3 classes ...fpcalico ...fpovs ...fplb\nin the calico case the bridge name field does not make senese to include as no bridges would be used.\n\nthat said a better solution would be possible if the neutron changes to pass a serislaed os_vif object where complete. when that is done you would  have 2 classes, 1 for the bridge based backends(ovs and linux bridge) and a second for routed backends that dose not contain the bridge_name field. on the neutron side you would pass two things back to nova the the os_vif vif object and the plugin to use to plug the interfaces.","commit_id":"0d237c70b5fd444f8b446854b35c61028c04f6e4"},{"author":{"_account_id":1779,"name":"Daniel Berrange","email":"berrange@redhat.com","username":"berrange"},"change_message_id":"9a1ec932c1dc44db50dc65531f57b2205d302fb6","unresolved":false,"context_lines":[{"line_number":140,"context_line":"        \u0027bridge_name\u0027: fields.StringField(),"},{"line_number":141,"context_line":""},{"line_number":142,"context_line":"        # MAC address of fp interface"},{"line_number":143,"context_line":"        \u0027tap_mac_address\u0027: fields.StringField(),"},{"line_number":144,"context_line":"    }"},{"line_number":145,"context_line":""},{"line_number":146,"context_line":""}],"source_content_type":"text/x-python","patch_set":1,"id":"5a8bc5a2_bf0a931f","line":143,"updated":"2016-09-23 14:54:17.000000000","message":"This seems dubious looking at the nova vif.py we only use a single MAC address, and we alread have that recorded in the parent class","commit_id":"0d237c70b5fd444f8b446854b35c61028c04f6e4"},{"author":{"_account_id":1779,"name":"Daniel Berrange","email":"berrange@redhat.com","username":"berrange"},"change_message_id":"d1f46f94c8d288018540a817b479184d12352221","unresolved":false,"context_lines":[{"line_number":140,"context_line":"        \u0027bridge_name\u0027: fields.StringField(),"},{"line_number":141,"context_line":""},{"line_number":142,"context_line":"        # MAC address of fp interface"},{"line_number":143,"context_line":"        \u0027tap_mac_address\u0027: fields.StringField(),"},{"line_number":144,"context_line":"    }"},{"line_number":145,"context_line":""},{"line_number":146,"context_line":""}],"source_content_type":"text/x-python","patch_set":1,"id":"5a8bc5a2_a28c3836","line":143,"in_reply_to":"5a8bc5a2_7f056b78","updated":"2016-09-23 15:22:23.000000000","message":"That\u0027s not currently used by anything except the \u0027TAP\u0027 vif type in nova","commit_id":"0d237c70b5fd444f8b446854b35c61028c04f6e4"},{"author":{"_account_id":17671,"name":"Francesco Santoro","email":"fraroot@gmail.com","username":"francesco.santoro"},"change_message_id":"ccdca5ed35c247c7f5d96912ccfbf9da9f0cdae6","unresolved":false,"context_lines":[{"line_number":140,"context_line":"        \u0027bridge_name\u0027: fields.StringField(),"},{"line_number":141,"context_line":""},{"line_number":142,"context_line":"        # MAC address of fp interface"},{"line_number":143,"context_line":"        \u0027tap_mac_address\u0027: fields.StringField(),"},{"line_number":144,"context_line":"    }"},{"line_number":145,"context_line":""},{"line_number":146,"context_line":""}],"source_content_type":"text/x-python","patch_set":1,"id":"5a8bc5a2_a8789790","line":143,"in_reply_to":"5a8bc5a2_a28c3836","updated":"2016-09-23 15:48:38.000000000","message":"Future developments for networking_6wind mechanism driver will be able to set the VIF_DETAILS_TAP_MAC_ADDRESS to support vhostuser plug with calico. The field I added is just to prepare such scenario.","commit_id":"0d237c70b5fd444f8b446854b35c61028c04f6e4"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"ed269c2b430150ee3282d16d0c9f35b30ce4270a","unresolved":false,"context_lines":[{"line_number":140,"context_line":"        \u0027bridge_name\u0027: fields.StringField(),"},{"line_number":141,"context_line":""},{"line_number":142,"context_line":"        # MAC address of fp interface"},{"line_number":143,"context_line":"        \u0027tap_mac_address\u0027: fields.StringField(),"},{"line_number":144,"context_line":"    }"},{"line_number":145,"context_line":""},{"line_number":146,"context_line":""}],"source_content_type":"text/x-python","patch_set":1,"id":"5a8bc5a2_f8612806","line":143,"in_reply_to":"5a8bc5a2_a8789790","updated":"2016-09-27 01:07:16.000000000","message":"in this case am i correct in saying that your intention is for nova in the future read the tap_mac_address fields and use that to set the mac adress of the tap it create for example by seting it in the libvirt xml?\n\nperhaps the addtion of this field should be in a seperate change that depends on the networking_6wind change.","commit_id":"0d237c70b5fd444f8b446854b35c61028c04f6e4"},{"author":{"_account_id":17671,"name":"Francesco Santoro","email":"fraroot@gmail.com","username":"francesco.santoro"},"change_message_id":"30d2fdb10e2c633b61bc3afaccbf7cd27e2d8f30","unresolved":false,"context_lines":[{"line_number":140,"context_line":"        \u0027bridge_name\u0027: fields.StringField(),"},{"line_number":141,"context_line":""},{"line_number":142,"context_line":"        # MAC address of fp interface"},{"line_number":143,"context_line":"        \u0027tap_mac_address\u0027: fields.StringField(),"},{"line_number":144,"context_line":"    }"},{"line_number":145,"context_line":""},{"line_number":146,"context_line":""}],"source_content_type":"text/x-python","patch_set":1,"id":"5a8bc5a2_7f056b78","line":143,"in_reply_to":"5a8bc5a2_bf0a931f","updated":"2016-09-23 15:12:15.000000000","message":"Here I mean the mac address that can be set by mechanism drivers via the VIF_DETAILS_TAP_MAC_ADDRESS field.","commit_id":"0d237c70b5fd444f8b446854b35c61028c04f6e4"},{"author":{"_account_id":17671,"name":"Francesco Santoro","email":"fraroot@gmail.com","username":"francesco.santoro"},"change_message_id":"db64466ff61a7860e8962573d8b7458368b27cfb","unresolved":false,"context_lines":[{"line_number":140,"context_line":"        \u0027bridge_name\u0027: fields.StringField(),"},{"line_number":141,"context_line":""},{"line_number":142,"context_line":"        # MAC address of fp interface"},{"line_number":143,"context_line":"        \u0027tap_mac_address\u0027: fields.StringField(),"},{"line_number":144,"context_line":"    }"},{"line_number":145,"context_line":""},{"line_number":146,"context_line":""}],"source_content_type":"text/x-python","patch_set":1,"id":"5a8bc5a2_95c37337","line":143,"in_reply_to":"5a8bc5a2_f8612806","updated":"2016-09-27 12:05:53.000000000","message":"Here the idea is to be able to change the mac address of the fastpath vhostuser port on the compute node using the TAP_MAC_ADDRESS to allow calico routing.\n\nI totally agree that this should live in a separate change.","commit_id":"0d237c70b5fd444f8b446854b35c61028c04f6e4"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"bf01f5a32b030255385c7bdafbb3f01e899df7aa","unresolved":false,"context_lines":[{"line_number":130,"context_line":""},{"line_number":131,"context_line":"    VERSION \u003d \u00271.0\u0027"},{"line_number":132,"context_line":""},{"line_number":133,"context_line":"    fields \u003d {"},{"line_number":134,"context_line":"        # Name of the fastpath device to create"},{"line_number":135,"context_line":"        \u0027vif_name\u0027: fields.StringField()"},{"line_number":136,"context_line":"    }"},{"line_number":137,"context_line":""},{"line_number":138,"context_line":""},{"line_number":139,"context_line":"@base.VersionedObjectRegistry.register"}],"source_content_type":"text/x-python","patch_set":2,"id":"da6895a0_ac6d4406","line":136,"range":{"start_line":133,"start_character":3,"end_line":136,"end_character":5},"updated":"2016-10-22 14:03:26.000000000","message":"this field could actully be aded to the existing VIFVHostUser.\n\nthis would allow use to pass it from nova/neutron instead of generate then same name again as we do currently.\n\nso if this is the only addtionaly field you need i would skip creating a new vif type for FP","commit_id":"bc0dd254c5f6221ce10cf59efd4b260139eac4ed"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"bf01f5a32b030255385c7bdafbb3f01e899df7aa","unresolved":false,"context_lines":[{"line_number":181,"context_line":"    VERSION \u003d \u00271.0\u0027"},{"line_number":182,"context_line":""},{"line_number":183,"context_line":"    fields \u003d {"},{"line_number":184,"context_line":"        \u0027manager_name\u0027: fields.StringField(),"},{"line_number":185,"context_line":"        \u0027hybrid_plug\u0027: fields.BooleanField(default\u003dFalse),"},{"line_number":186,"context_line":"    }"},{"line_number":187,"context_line":""}],"source_content_type":"text/x-python","patch_set":2,"id":"da6895a0_2c821495","line":184,"range":{"start_line":184,"start_character":7,"end_line":184,"end_character":45},"updated":"2016-10-22 14:03:26.000000000","message":"what are the posible valudes of manager_name? what is it for.","commit_id":"bc0dd254c5f6221ce10cf59efd4b260139eac4ed"},{"author":{"_account_id":17671,"name":"Francesco Santoro","email":"fraroot@gmail.com","username":"francesco.santoro"},"change_message_id":"e4996ad648e006c2af6ca4ad6b2cf9101831fd85","unresolved":false,"context_lines":[{"line_number":181,"context_line":"    VERSION \u003d \u00271.0\u0027"},{"line_number":182,"context_line":""},{"line_number":183,"context_line":"    fields \u003d {"},{"line_number":184,"context_line":"        \u0027manager_name\u0027: fields.StringField(),"},{"line_number":185,"context_line":"        \u0027hybrid_plug\u0027: fields.BooleanField(default\u003dFalse),"},{"line_number":186,"context_line":"    }"},{"line_number":187,"context_line":""}],"source_content_type":"text/x-python","patch_set":2,"id":"da6895a0_dbc4e360","line":184,"range":{"start_line":184,"start_character":7,"end_line":184,"end_character":45},"in_reply_to":"da6895a0_2c821495","updated":"2016-10-24 07:46:34.000000000","message":"It is actually the equivalent of bridge_name for VIFBridge types. I chose manager_name to avoid confusion.\nIf it is ok for you I will use bridge_name or port_bridge.","commit_id":"bc0dd254c5f6221ce10cf59efd4b260139eac4ed"},{"author":{"_account_id":1779,"name":"Daniel Berrange","email":"berrange@redhat.com","username":"berrange"},"change_message_id":"774ca4e78f84854da2ad49eae63b497e4a5a96c0","unresolved":false,"context_lines":[{"line_number":200,"context_line":"    fields \u003d {"},{"line_number":201,"context_line":"        # The mac address of the host vhostuser port"},{"line_number":202,"context_line":"        \u0027mac_address\u0027: fields.MACAddressField(nullable\u003dTrue),"},{"line_number":203,"context_line":"    }"},{"line_number":204,"context_line":""},{"line_number":205,"context_line":""},{"line_number":206,"context_line":"@base.VersionedObjectRegistry.register"}],"source_content_type":"text/x-python","patch_set":4,"id":"ba5da102_6026cacf","line":203,"updated":"2016-10-28 13:01:35.000000000","message":"I\u0027m not sure why you\u0027ve done this all as port profiles. They should be VIF type objects. Port profile object info will not involved in the nova/neutron config negotiation, not is is exposed by plugins declaring their supoported VIFs. So by using port profiles you make discovery of FP support impossible with the current design.","commit_id":"71616c00354faac067e8000ac9056adc4cb8a29a"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"57f520d72912d92e3c95e0df8130927f66445ce5","unresolved":false,"context_lines":[{"line_number":200,"context_line":"    fields \u003d {"},{"line_number":201,"context_line":"        # The mac address of the host vhostuser port"},{"line_number":202,"context_line":"        \u0027mac_address\u0027: fields.MACAddressField(nullable\u003dTrue),"},{"line_number":203,"context_line":"    }"},{"line_number":204,"context_line":""},{"line_number":205,"context_line":""},{"line_number":206,"context_line":"@base.VersionedObjectRegistry.register"}],"source_content_type":"text/x-python","patch_set":4,"id":"ba5da102_58c60b4a","line":203,"in_reply_to":"ba5da102_6026cacf","updated":"2016-10-29 19:24:43.000000000","message":"the negotiation between nova and neutron has not been defined yet. we will need a spec in nova/neutron to trash out the details but the initial version that you were proposing is not general enough to convey all required information from nova to neutron to complete a robost binding, so I don\u0027t see why we can\u0027t pass both vif and port profiles they support.  \n\nfor example by just passing vif we have no way of say which of the in tree vifs are available, after all since they are in tree ovs and linux bridge will always be available even on windows. \n\nsimilarly, the existing vif_vhostuser indicate that we can use a vhost user interface but it does not tell us if we can use it with qemu as the server or as the client.","commit_id":"71616c00354faac067e8000ac9056adc4cb8a29a"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"3722f5c13540e0a23e8378ca66da5c4a8bd51b5e","unresolved":false,"context_lines":[{"line_number":143,"context_line":""},{"line_number":144,"context_line":""},{"line_number":145,"context_line":"@base.VersionedObjectRegistry.register"},{"line_number":146,"context_line":"class VIFVHostUserFP(VIFVHostUser):"},{"line_number":147,"context_line":"    # For libvirt drivers, this maps to type\u003d\u0027vhostuser\u0027"},{"line_number":148,"context_line":""},{"line_number":149,"context_line":"    VERSION \u003d \u00271.0\u0027"},{"line_number":150,"context_line":""},{"line_number":151,"context_line":""},{"line_number":152,"context_line":"@base.VersionedObjectRegistry.register"},{"line_number":153,"context_line":"class VIFHostDevice(VIFBase):"},{"line_number":154,"context_line":"    # For libvirt drivers, this maps to type\u003d\u0027hostdev\u0027"},{"line_number":155,"context_line":""}],"source_content_type":"text/x-python","patch_set":11,"id":"1a6eadb0_eebbd00c","line":152,"range":{"start_line":146,"start_character":0,"end_line":152,"end_character":38},"updated":"2016-12-13 01:29:18.000000000","message":"this change is not required and should be removed.","commit_id":"8847c488d057db67a158150313f01dadfc05fba6"},{"author":{"_account_id":17671,"name":"Francesco Santoro","email":"fraroot@gmail.com","username":"francesco.santoro"},"change_message_id":"7022ec2021a97b4192475694c5fc2f73913faccb","unresolved":false,"context_lines":[{"line_number":143,"context_line":""},{"line_number":144,"context_line":""},{"line_number":145,"context_line":"@base.VersionedObjectRegistry.register"},{"line_number":146,"context_line":"class VIFVHostUserFP(VIFVHostUser):"},{"line_number":147,"context_line":"    # For libvirt drivers, this maps to type\u003d\u0027vhostuser\u0027"},{"line_number":148,"context_line":""},{"line_number":149,"context_line":"    VERSION \u003d \u00271.0\u0027"},{"line_number":150,"context_line":""},{"line_number":151,"context_line":""},{"line_number":152,"context_line":"@base.VersionedObjectRegistry.register"},{"line_number":153,"context_line":"class VIFHostDevice(VIFBase):"},{"line_number":154,"context_line":"    # For libvirt drivers, this maps to type\u003d\u0027hostdev\u0027"},{"line_number":155,"context_line":""}],"source_content_type":"text/x-python","patch_set":11,"id":"1a6eadb0_6b524515","line":152,"range":{"start_line":146,"start_character":0,"end_line":152,"end_character":38},"in_reply_to":"1a6eadb0_eebbd00c","updated":"2016-12-13 09:09:31.000000000","message":"Ok no problem. I kept the class definition because I though it could help during negotiation. But if the approach is to manage fast path detection using both vif type and port profiles there is no reason to keep this class. Get rid of this class will also allow to keep only one _set_config_VIFVHostUser method on nova side.","commit_id":"8847c488d057db67a158150313f01dadfc05fba6"}],"os_vif/tests/test_vif.py":[{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"bf01f5a32b030255385c7bdafbb3f01e899df7aa","unresolved":false,"context_lines":[{"line_number":97,"context_line":"        prof \u003d objects.vif.VIFPortProfileFPOpenVSwitch("},{"line_number":98,"context_line":"            interface_id\u003d\"07bd6cea-fb37-4594-b769-90fc51854ee8\","},{"line_number":99,"context_line":"            profile_id\u003d\"fishfood\","},{"line_number":100,"context_line":"            manager_name\u003d\"br-int\","},{"line_number":101,"context_line":"            hybrid_plug\u003dFalse)"},{"line_number":102,"context_line":"        self._test_vif(objects.vif.VIFVHostUserFP,"},{"line_number":103,"context_line":"                       path\u003d\"/some/socket.path\","}],"source_content_type":"text/x-python","patch_set":2,"id":"da6895a0_6c4a4c85","line":100,"range":{"start_line":100,"start_character":12,"end_line":100,"end_character":34},"updated":"2016-10-22 14:03:26.000000000","message":"why call it manager_name if it is a bridge?\nif it would only ever be the name of the integration bridge then it would be better to call it int_br as is done in neutron","commit_id":"bc0dd254c5f6221ce10cf59efd4b260139eac4ed"},{"author":{"_account_id":17671,"name":"Francesco Santoro","email":"fraroot@gmail.com","username":"francesco.santoro"},"change_message_id":"e4996ad648e006c2af6ca4ad6b2cf9101831fd85","unresolved":false,"context_lines":[{"line_number":97,"context_line":"        prof \u003d objects.vif.VIFPortProfileFPOpenVSwitch("},{"line_number":98,"context_line":"            interface_id\u003d\"07bd6cea-fb37-4594-b769-90fc51854ee8\","},{"line_number":99,"context_line":"            profile_id\u003d\"fishfood\","},{"line_number":100,"context_line":"            manager_name\u003d\"br-int\","},{"line_number":101,"context_line":"            hybrid_plug\u003dFalse)"},{"line_number":102,"context_line":"        self._test_vif(objects.vif.VIFVHostUserFP,"},{"line_number":103,"context_line":"                       path\u003d\"/some/socket.path\","}],"source_content_type":"text/x-python","patch_set":2,"id":"da6895a0_1b8c9b34","line":100,"range":{"start_line":100,"start_character":12,"end_line":100,"end_character":34},"in_reply_to":"da6895a0_6c4a4c85","updated":"2016-10-24 07:46:34.000000000","message":"It could be a linuxbridge name if hybrid plug is active. But yes, this manager name is the name of the bridge that connects fastpath ports.","commit_id":"bc0dd254c5f6221ce10cf59efd4b260139eac4ed"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"39e4665cf011f57f52af9ee1c920f61c3fe0c27e","unresolved":false,"context_lines":[{"line_number":102,"context_line":"            hybrid_plug\u003dFalse)"},{"line_number":103,"context_line":"        self._test_vif(objects.vif.VIFVHostUser,"},{"line_number":104,"context_line":"                       path\u003d\"/some/socket.path\","},{"line_number":105,"context_line":"                       mode\u003dobjects.fields.VIFVHostUserMode.CLIENT,"},{"line_number":106,"context_line":"                       vif_name\u003d\"tap123\","},{"line_number":107,"context_line":"                       port_profile\u003dprof)"},{"line_number":108,"context_line":""}],"source_content_type":"text/x-python","patch_set":3,"id":"da6895a0_2bef23b1","line":105,"range":{"start_line":105,"start_character":22,"end_line":105,"end_character":67},"updated":"2016-10-25 19:06:41.000000000","message":"by the way does 6winds fast path stack support both client and server mode? i know vpp does and support was recently added to ovs with dpdk but i have not looked into fast path too closely.","commit_id":"b962c6609ff1ad83537557d94f5c209a7346c8eb"},{"author":{"_account_id":17671,"name":"Francesco Santoro","email":"fraroot@gmail.com","username":"francesco.santoro"},"change_message_id":"fa8573ce01606b3665099b19aa958363ed0f2f60","unresolved":false,"context_lines":[{"line_number":102,"context_line":"            hybrid_plug\u003dFalse)"},{"line_number":103,"context_line":"        self._test_vif(objects.vif.VIFVHostUser,"},{"line_number":104,"context_line":"                       path\u003d\"/some/socket.path\","},{"line_number":105,"context_line":"                       mode\u003dobjects.fields.VIFVHostUserMode.CLIENT,"},{"line_number":106,"context_line":"                       vif_name\u003d\"tap123\","},{"line_number":107,"context_line":"                       port_profile\u003dprof)"},{"line_number":108,"context_line":""}],"source_content_type":"text/x-python","patch_set":3,"id":"ba5da102_c778c892","line":105,"range":{"start_line":105,"start_character":22,"end_line":105,"end_character":67},"in_reply_to":"da6895a0_2bef23b1","updated":"2016-10-26 09:25:10.000000000","message":"fast path supports both modes. But we use CLIENT as default case.","commit_id":"b962c6609ff1ad83537557d94f5c209a7346c8eb"}]}
