)]}'
{"/COMMIT_MSG":[{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"bc9a0ecf42bd11c0f59764191ce3e6251b7c490e","unresolved":false,"context_lines":[{"line_number":7,"context_line":"Revert \"Set device_owner for trunk subports\""},{"line_number":8,"context_line":""},{"line_number":9,"context_line":"This is not required by business logic as pointed out in the commit"},{"line_number":10,"context_line":"message, and can lead to race problems. For instance, if a user wants to"},{"line_number":11,"context_line":"set a different device_owner for the subport and triggers an update_port"},{"line_number":12,"context_line":"call to do that right after calling trunk_add_subport, sometimes"},{"line_number":13,"context_line":"(actually most of the time) the follow up update port call may happen"},{"line_number":14,"context_line":"before the internal update_port call made by trunk_add_subport."},{"line_number":15,"context_line":"Consequently, the device_owner may end up with the wrong value, in this"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":3,"id":"ba5201f7_4ad4541f","line":12,"range":{"start_line":10,"start_character":54,"end_line":12,"end_character":53},"updated":"2017-01-11 16:54:18.000000000","message":"Can you elaborate as to why a user would want to do that. I\u0027d say this practice should be strongly discouraged as it may lead to other unknown side effects.","commit_id":"b55ed651ceb1695e9a78c9d6b57db0e8c98059b2"},{"author":{"_account_id":23567,"name":"Luis Tomas Bolivar","email":"ltomasbo@redhat.com","username":"ltomasbo"},"change_message_id":"ccce37eaea7ff30b6156ca5a13f9138d87e6dfcf","unresolved":false,"context_lines":[{"line_number":7,"context_line":"Revert \"Set device_owner for trunk subports\""},{"line_number":8,"context_line":""},{"line_number":9,"context_line":"This is not required by business logic as pointed out in the commit"},{"line_number":10,"context_line":"message, and can lead to race problems. For instance, if a user wants to"},{"line_number":11,"context_line":"set a different device_owner for the subport and triggers an update_port"},{"line_number":12,"context_line":"call to do that right after calling trunk_add_subport, sometimes"},{"line_number":13,"context_line":"(actually most of the time) the follow up update port call may happen"},{"line_number":14,"context_line":"before the internal update_port call made by trunk_add_subport."},{"line_number":15,"context_line":"Consequently, the device_owner may end up with the wrong value, in this"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":3,"id":"5a3905b3_831b69c9","line":12,"range":{"start_line":10,"start_character":54,"end_line":12,"end_character":53},"in_reply_to":"5a3905b3_66dec240","updated":"2017-01-23 10:40:31.000000000","message":"No problem! Thanks for taking the time to discuss this!\n\nYes, we are using device_owner as a filter mechanism, but also to highlight that those subports are actually created by the kuryr service. Anyway, I think moving to tagging is a good idea, although not available yet, right?","commit_id":"b55ed651ceb1695e9a78c9d6b57db0e8c98059b2"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"e5d7e35ffe28c3d5a03d0eb754650f584ec31181","unresolved":false,"context_lines":[{"line_number":7,"context_line":"Revert \"Set device_owner for trunk subports\""},{"line_number":8,"context_line":""},{"line_number":9,"context_line":"This is not required by business logic as pointed out in the commit"},{"line_number":10,"context_line":"message, and can lead to race problems. For instance, if a user wants to"},{"line_number":11,"context_line":"set a different device_owner for the subport and triggers an update_port"},{"line_number":12,"context_line":"call to do that right after calling trunk_add_subport, sometimes"},{"line_number":13,"context_line":"(actually most of the time) the follow up update port call may happen"},{"line_number":14,"context_line":"before the internal update_port call made by trunk_add_subport."},{"line_number":15,"context_line":"Consequently, the device_owner may end up with the wrong value, in this"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":3,"id":"5a3905b3_942c659a","line":12,"range":{"start_line":10,"start_character":54,"end_line":12,"end_character":53},"in_reply_to":"5a3905b3_831b69c9","updated":"2017-01-23 19:51:49.000000000","message":"I hope we can make it available as Ocata ships. You should track https://review.openstack.org/#/c/413662 to see when it lands.","commit_id":"b55ed651ceb1695e9a78c9d6b57db0e8c98059b2"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"390e3b90c788b88292ddea4ff9330789b654cb70","unresolved":false,"context_lines":[{"line_number":7,"context_line":"Revert \"Set device_owner for trunk subports\""},{"line_number":8,"context_line":""},{"line_number":9,"context_line":"This is not required by business logic as pointed out in the commit"},{"line_number":10,"context_line":"message, and can lead to race problems. For instance, if a user wants to"},{"line_number":11,"context_line":"set a different device_owner for the subport and triggers an update_port"},{"line_number":12,"context_line":"call to do that right after calling trunk_add_subport, sometimes"},{"line_number":13,"context_line":"(actually most of the time) the follow up update port call may happen"},{"line_number":14,"context_line":"before the internal update_port call made by trunk_add_subport."},{"line_number":15,"context_line":"Consequently, the device_owner may end up with the wrong value, in this"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":3,"id":"da36d5c6_05c32539","line":12,"range":{"start_line":10,"start_character":54,"end_line":12,"end_character":53},"in_reply_to":"5a3905b3_942c659a","updated":"2017-02-22 21:11:45.000000000","message":"@Luis: btw the patch merged for Ocata so if you\u0027re ok with it you may no longer need this one.","commit_id":"b55ed651ceb1695e9a78c9d6b57db0e8c98059b2"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"0d375e58654938de1db6b05ae270caa8beb5025d","unresolved":false,"context_lines":[{"line_number":7,"context_line":"Revert \"Set device_owner for trunk subports\""},{"line_number":8,"context_line":""},{"line_number":9,"context_line":"This is not required by business logic as pointed out in the commit"},{"line_number":10,"context_line":"message, and can lead to race problems. For instance, if a user wants to"},{"line_number":11,"context_line":"set a different device_owner for the subport and triggers an update_port"},{"line_number":12,"context_line":"call to do that right after calling trunk_add_subport, sometimes"},{"line_number":13,"context_line":"(actually most of the time) the follow up update port call may happen"},{"line_number":14,"context_line":"before the internal update_port call made by trunk_add_subport."},{"line_number":15,"context_line":"Consequently, the device_owner may end up with the wrong value, in this"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":3,"id":"5a3905b3_66dec240","line":12,"range":{"start_line":10,"start_character":54,"end_line":12,"end_character":53},"in_reply_to":"9a57fde8_722c36ac","updated":"2017-01-20 20:20:51.000000000","message":"Sorry for taking a while to get back here.\n\nAre you using the device_owner simply as a filter mechanism? If so, would you be open to using tagging instead [1]? \n\nEven though device_owner is used in a few places by ironic and nova, the handling of device_owner has given us a bit of grief in the past and if I could back I\u0027d not expose the field at all. Right now the device_id is blank also, and that is also used by services like nova to store the VM-uuid.\n\nI suppose we can either proceed with the revert here (but we should clear any DB record of the device_owner before proceeding) or let you change approach in kuryr and use tagging instead. I\u0027ll pull Kevin into the discussion, as I am sure he\u0027ll have a good perspective on it.\n\n[1] https://review.openstack.org/#/c/413662","commit_id":"b55ed651ceb1695e9a78c9d6b57db0e8c98059b2"},{"author":{"_account_id":23567,"name":"Luis Tomas Bolivar","email":"ltomasbo@redhat.com","username":"ltomasbo"},"change_message_id":"a348b87303924101ebbc7dd47460b427e848534a","unresolved":false,"context_lines":[{"line_number":7,"context_line":"Revert \"Set device_owner for trunk subports\""},{"line_number":8,"context_line":""},{"line_number":9,"context_line":"This is not required by business logic as pointed out in the commit"},{"line_number":10,"context_line":"message, and can lead to race problems. For instance, if a user wants to"},{"line_number":11,"context_line":"set a different device_owner for the subport and triggers an update_port"},{"line_number":12,"context_line":"call to do that right after calling trunk_add_subport, sometimes"},{"line_number":13,"context_line":"(actually most of the time) the follow up update port call may happen"},{"line_number":14,"context_line":"before the internal update_port call made by trunk_add_subport."},{"line_number":15,"context_line":"Consequently, the device_owner may end up with the wrong value, in this"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":3,"id":"ba5201f7_bd142844","line":12,"range":{"start_line":10,"start_character":54,"end_line":12,"end_character":53},"in_reply_to":"ba5201f7_4ad4541f","updated":"2017-01-11 17:09:03.000000000","message":"I\u0027m hitting this problem when using Trunk Ports together with Kuryr. We have different VLANs (subports) for each container running inside the VM. Then, all the subports are created by kuryr service and device_owner needs to be set by kuryr (as the containers are calling kuryr driver that in turn calls neutron to create the ports). Therefore, after attaching the subport to the trunk, we need to call update_port to set the device_owner (it would be great to be able to pass it to trunk_add_subport directly, but that would be a bigger change)","commit_id":"b55ed651ceb1695e9a78c9d6b57db0e8c98059b2"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"c3ab6a514064be975b6778d1d1205f2a3dbeb6fa","unresolved":false,"context_lines":[{"line_number":7,"context_line":"Revert \"Set device_owner for trunk subports\""},{"line_number":8,"context_line":""},{"line_number":9,"context_line":"This is not required by business logic as pointed out in the commit"},{"line_number":10,"context_line":"message, and can lead to race problems. For instance, if a user wants to"},{"line_number":11,"context_line":"set a different device_owner for the subport and triggers an update_port"},{"line_number":12,"context_line":"call to do that right after calling trunk_add_subport, sometimes"},{"line_number":13,"context_line":"(actually most of the time) the follow up update port call may happen"},{"line_number":14,"context_line":"before the internal update_port call made by trunk_add_subport."},{"line_number":15,"context_line":"Consequently, the device_owner may end up with the wrong value, in this"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":3,"id":"ba5201f7_ccb695f8","line":12,"range":{"start_line":10,"start_character":54,"end_line":12,"end_character":53},"in_reply_to":"ba5201f7_bd142844","updated":"2017-01-11 21:02:04.000000000","message":"Interesting, I would definitely love to make sure kuryr works well with this use case, but what is the exact workflow? I would like to have a good picture before making a judgement call.","commit_id":"b55ed651ceb1695e9a78c9d6b57db0e8c98059b2"},{"author":{"_account_id":23567,"name":"Luis Tomas Bolivar","email":"ltomasbo@redhat.com","username":"ltomasbo"},"change_message_id":"6f6663c4794f7e47a7c37b53eb1f3a53c89a60e1","unresolved":false,"context_lines":[{"line_number":7,"context_line":"Revert \"Set device_owner for trunk subports\""},{"line_number":8,"context_line":""},{"line_number":9,"context_line":"This is not required by business logic as pointed out in the commit"},{"line_number":10,"context_line":"message, and can lead to race problems. For instance, if a user wants to"},{"line_number":11,"context_line":"set a different device_owner for the subport and triggers an update_port"},{"line_number":12,"context_line":"call to do that right after calling trunk_add_subport, sometimes"},{"line_number":13,"context_line":"(actually most of the time) the follow up update port call may happen"},{"line_number":14,"context_line":"before the internal update_port call made by trunk_add_subport."},{"line_number":15,"context_line":"Consequently, the device_owner may end up with the wrong value, in this"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":3,"id":"9a57fde8_722c36ac","line":12,"range":{"start_line":10,"start_character":54,"end_line":12,"end_character":53},"in_reply_to":"ba5201f7_ccb695f8","updated":"2017-01-12 08:29:06.000000000","message":"The workflow is the next (let\u0027s focus on raw docker for simplicity):\n- Create a VM with Trunk Ports (parent)\n- Install kuryr service inside the VM\n- (inside the VM) With docker, create a network by using the kuryr driver (kuryr will make the calls to neutron to create the networks)\n- (inside the VM) Boot a container in that network, which entails kuryr service doing the next actions:\n  1) Create a neutron port\n  2) Attach neutron port to the trunk (i.e., subport)\n  3) Change the neutron port device_owner to belong to kuryr-service as it is the entity in charge of managing (creating/removing/updating) these ports)\n  4) Boot the container (this binds the contaoiner to the VM vNIC in the proper vlan","commit_id":"b55ed651ceb1695e9a78c9d6b57db0e8c98059b2"},{"author":{"_account_id":23567,"name":"Luis Tomas Bolivar","email":"ltomasbo@redhat.com","username":"ltomasbo"},"change_message_id":"125855002e05c4d7fc070a2f0011412dfacd2741","unresolved":false,"context_lines":[{"line_number":7,"context_line":"Revert \"Set device_owner for trunk subports\""},{"line_number":8,"context_line":""},{"line_number":9,"context_line":"This is not required by business logic as pointed out in the commit"},{"line_number":10,"context_line":"message, and can lead to race problems. For instance, if a user wants to"},{"line_number":11,"context_line":"set a different device_owner for the subport and triggers an update_port"},{"line_number":12,"context_line":"call to do that right after calling trunk_add_subport, sometimes"},{"line_number":13,"context_line":"(actually most of the time) the follow up update port call may happen"},{"line_number":14,"context_line":"before the internal update_port call made by trunk_add_subport."},{"line_number":15,"context_line":"Consequently, the device_owner may end up with the wrong value, in this"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":3,"id":"da36d5c6_f07163a9","line":12,"range":{"start_line":10,"start_character":54,"end_line":12,"end_character":53},"in_reply_to":"da36d5c6_05c32539","updated":"2017-02-23 07:50:54.000000000","message":"Ok, works for me, I can close this. Actually we already merged the changes to move from device_owner to tags in Kuryr to avoid collisions. That said, we still keep storing the value compute:kuryr in the device_owner for the ports created by kuryr as this follows the (most accepted) format for that field.","commit_id":"b55ed651ceb1695e9a78c9d6b57db0e8c98059b2"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"bc9a0ecf42bd11c0f59764191ce3e6251b7c490e","unresolved":false,"context_lines":[{"line_number":15,"context_line":"Consequently, the device_owner may end up with the wrong value, in this"},{"line_number":16,"context_line":"case the hardcoded trunk:subport."},{"line_number":17,"context_line":""},{"line_number":18,"context_line":"This reverts commit 72bc7551e8af05ebf9bdab560eeced41823d9502."},{"line_number":19,"context_line":""},{"line_number":20,"context_line":"Change-Id: I5a66f7956dfb2fbc6e6a184540ef320307c24dd8"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":3,"id":"ba5201f7_0a9dec65","line":18,"updated":"2017-01-11 16:54:18.000000000","message":"We can\u0027t simply revert this commit, by doing so, existing deployments that may have trunks in use would need to be \u0027sanitized\u0027. That said, I would like you to elaborate more on the use case you\u0027re trying to achieve.","commit_id":"b55ed651ceb1695e9a78c9d6b57db0e8c98059b2"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"4f576b38dab35353936fbb1a90c942619dd58362","unresolved":false,"context_lines":[{"line_number":15,"context_line":"Consequently, the device_owner may end up with the wrong value, in this"},{"line_number":16,"context_line":"case the hardcoded trunk:subport."},{"line_number":17,"context_line":""},{"line_number":18,"context_line":"This reverts commit 72bc7551e8af05ebf9bdab560eeced41823d9502."},{"line_number":19,"context_line":""},{"line_number":20,"context_line":"Change-Id: I5a66f7956dfb2fbc6e6a184540ef320307c24dd8"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":3,"id":"ba5201f7_8a09dc86","line":18,"in_reply_to":"ba5201f7_0a9dec65","updated":"2017-01-11 16:56:26.000000000","message":"...before proceeding.","commit_id":"b55ed651ceb1695e9a78c9d6b57db0e8c98059b2"},{"author":{"_account_id":23567,"name":"Luis Tomas Bolivar","email":"ltomasbo@redhat.com","username":"ltomasbo"},"change_message_id":"a348b87303924101ebbc7dd47460b427e848534a","unresolved":false,"context_lines":[{"line_number":15,"context_line":"Consequently, the device_owner may end up with the wrong value, in this"},{"line_number":16,"context_line":"case the hardcoded trunk:subport."},{"line_number":17,"context_line":""},{"line_number":18,"context_line":"This reverts commit 72bc7551e8af05ebf9bdab560eeced41823d9502."},{"line_number":19,"context_line":""},{"line_number":20,"context_line":"Change-Id: I5a66f7956dfb2fbc6e6a184540ef320307c24dd8"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":3,"id":"ba5201f7_1dec54f2","line":18,"in_reply_to":"ba5201f7_0a9dec65","updated":"2017-01-11 17:09:03.000000000","message":"True.\n\nWould it be ok to include an extra check about the value of TRUNK_SUBPORT_OWNER, and if it is set to \u0027none\u0027, don\u0027t include it on the call, and if it is set to the current string, just do what it does currently. If that so, I can write a new patch with that, instead of a revert.","commit_id":"b55ed651ceb1695e9a78c9d6b57db0e8c98059b2"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"c3ab6a514064be975b6778d1d1205f2a3dbeb6fa","unresolved":false,"context_lines":[{"line_number":15,"context_line":"Consequently, the device_owner may end up with the wrong value, in this"},{"line_number":16,"context_line":"case the hardcoded trunk:subport."},{"line_number":17,"context_line":""},{"line_number":18,"context_line":"This reverts commit 72bc7551e8af05ebf9bdab560eeced41823d9502."},{"line_number":19,"context_line":""},{"line_number":20,"context_line":"Change-Id: I5a66f7956dfb2fbc6e6a184540ef320307c24dd8"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":3,"id":"ba5201f7_ccadb59c","line":18,"in_reply_to":"ba5201f7_1dec54f2","updated":"2017-01-11 21:02:04.000000000","message":"let\u0027s elaborate on the use case a bit longer before coming up with a solution.","commit_id":"b55ed651ceb1695e9a78c9d6b57db0e8c98059b2"}]}
