)]}'
{"/COMMIT_MSG":[{"author":{"_account_id":9656,"name":"Ihar Hrachyshka","email":"ihrachys@redhat.com","username":"ihrachys","status":"Red Hat Networking Systems Engineer"},"change_message_id":"f21d2546b6b426b53775eb5a9e2aa808b27c226d","unresolved":false,"context_lines":[{"line_number":16,"context_line":"from Openvswitch monitor."},{"line_number":17,"context_line":"On fullstack tests however we can assume that agent\u0027s and server\u0027s"},{"line_number":18,"context_line":"config are in sync so trunk driver can be initialized only if"},{"line_number":19,"context_line":"\"trunk\" service plugin is enabled on server side."},{"line_number":20,"context_line":""},{"line_number":21,"context_line":"Change-Id: I3ad8d6e7b8f103867ee277078d03f3a01c20ac0d"},{"line_number":22,"context_line":"Closes-Bug: #1687709"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":1,"id":"3f4b6375_5022b6ba","line":19,"updated":"2017-11-03 15:12:27.000000000","message":"There is a big assumption being made here: that only a single set of trunk-enabled services will run at the same time. If you would have two test classes that both enable trunk service plugin, you would still hit the race condition.\n\nI guess we can go with the solution but we need to document (NOTE/TODO), in code, why we do that, what limitations are, what is the expect path forward (f.e. link to the bug/patch suggesting full isolation of services).","commit_id":"477b938cef174160b38d43dbdf6d54b96cc1e0bd"},{"author":{"_account_id":11975,"name":"Slawek Kaplonski","email":"skaplons@redhat.com","username":"slaweq"},"change_message_id":"e4801601ca47305357c4af2721178621675fa9ed","unresolved":false,"context_lines":[{"line_number":16,"context_line":"from Openvswitch monitor."},{"line_number":17,"context_line":"On fullstack tests however we can assume that agent\u0027s and server\u0027s"},{"line_number":18,"context_line":"config are in sync so trunk driver can be initialized only if"},{"line_number":19,"context_line":"\"trunk\" service plugin is enabled on server side."},{"line_number":20,"context_line":""},{"line_number":21,"context_line":"Change-Id: I3ad8d6e7b8f103867ee277078d03f3a01c20ac0d"},{"line_number":22,"context_line":"Closes-Bug: #1687709"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":1,"id":"3f4b6375_39e4c50c","line":19,"in_reply_to":"3f4b6375_5022b6ba","updated":"2017-11-03 16:50:28.000000000","message":"I added it in https://review.openstack.org/#/c/517598/2/neutron/tests/fullstack/test_trunk.py","commit_id":"477b938cef174160b38d43dbdf6d54b96cc1e0bd"}],"neutron/tests/fullstack/cmd/ovs_agent.py":[{"author":{"_account_id":9656,"name":"Ihar Hrachyshka","email":"ihrachys@redhat.com","username":"ihrachys","status":"Red Hat Networking Systems Engineer"},"change_message_id":"f21d2546b6b426b53775eb5a9e2aa808b27c226d","unresolved":false,"context_lines":[{"line_number":28,"context_line":"    def new_init_handler(resource, event, trigger, payload\u003dNone):"},{"line_number":29,"context_line":"        # make this setup conditional based on server-side capabilities"},{"line_number":30,"context_line":"        # for fullstack tests we can assume that server-side and agent-side"},{"line_number":31,"context_line":"        # conf are in sync"},{"line_number":32,"context_line":"        if \"trunk\" not in cfg.CONF.service_plugins:"},{"line_number":33,"context_line":"            return"},{"line_number":34,"context_line":"        original_handler(resource, event, trigger, payload)"}],"source_content_type":"text/x-python","patch_set":1,"id":"3f4b6375_500bd628","line":31,"updated":"2017-11-03 15:12:27.000000000","message":"please also add TODO with a link to Jakub\u0027s work and the bug","commit_id":"477b938cef174160b38d43dbdf6d54b96cc1e0bd"},{"author":{"_account_id":11975,"name":"Slawek Kaplonski","email":"skaplons@redhat.com","username":"slaweq"},"change_message_id":"e4801601ca47305357c4af2721178621675fa9ed","unresolved":false,"context_lines":[{"line_number":28,"context_line":"    def new_init_handler(resource, event, trigger, payload\u003dNone):"},{"line_number":29,"context_line":"        # make this setup conditional based on server-side capabilities"},{"line_number":30,"context_line":"        # for fullstack tests we can assume that server-side and agent-side"},{"line_number":31,"context_line":"        # conf are in sync"},{"line_number":32,"context_line":"        if \"trunk\" not in cfg.CONF.service_plugins:"},{"line_number":33,"context_line":"            return"},{"line_number":34,"context_line":"        original_handler(resource, event, trigger, payload)"}],"source_content_type":"text/x-python","patch_set":1,"id":"3f4b6375_19ed89ef","line":31,"in_reply_to":"3f4b6375_500bd628","updated":"2017-11-03 16:50:28.000000000","message":"Done","commit_id":"477b938cef174160b38d43dbdf6d54b96cc1e0bd"}],"neutron/tests/fullstack/test_trunk.py":[{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"c53c7ec31607d5df919c42b97f8e91facfc2a7b4","unresolved":false,"context_lines":[{"line_number":211,"context_line":"    # https://bugs.launchpad.net/neutron/+bug/1687709 when more than one"},{"line_number":212,"context_line":"    # different ovs-agent with enabled trunk driver is running at a time it"},{"line_number":213,"context_line":"    # might lead to race contitions between them."},{"line_number":214,"context_line":"    # Because of that ovs_agent used for fullstack tests is monkeypatched and"},{"line_number":215,"context_line":"    # loads trunk driver only if trunk service plugin is enabled."},{"line_number":216,"context_line":"    # That makes restriction that only a single set of tests with trunk-enabled"},{"line_number":217,"context_line":"    # services will run at the same time."},{"line_number":218,"context_line":"    def test_trunk_lifecycle(self):"}],"source_content_type":"text/x-python","patch_set":2,"id":"3f4b6375_f9d3cdff","line":215,"range":{"start_line":214,"start_character":6,"end_line":215,"end_character":64},"updated":"2017-11-03 17:00:34.000000000","message":"Once we were down the path of monkey-patching, why did you choose the route of disabling the trunk logic vs monkey-patching the method that generates the bridge name? The latter would not lead to this limitation. I must be missing something.","commit_id":"806cf71eb5af34359f8bc0c6b740752e53cf854f"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"a885191131cba2642d69e0ece64113677fd54440","unresolved":false,"context_lines":[{"line_number":211,"context_line":"    # https://bugs.launchpad.net/neutron/+bug/1687709 when more than one"},{"line_number":212,"context_line":"    # different ovs-agent with enabled trunk driver is running at a time it"},{"line_number":213,"context_line":"    # might lead to race contitions between them."},{"line_number":214,"context_line":"    # Because of that ovs_agent used for fullstack tests is monkeypatched and"},{"line_number":215,"context_line":"    # loads trunk driver only if trunk service plugin is enabled."},{"line_number":216,"context_line":"    # That makes restriction that only a single set of tests with trunk-enabled"},{"line_number":217,"context_line":"    # services will run at the same time."},{"line_number":218,"context_line":"    def test_trunk_lifecycle(self):"}],"source_content_type":"text/x-python","patch_set":2,"id":"3f4b6375_e44d6472","line":215,"range":{"start_line":214,"start_character":6,"end_line":215,"end_character":64},"in_reply_to":"3f4b6375_392285ba","updated":"2017-11-03 17:15:20.000000000","message":"I thought the amount of code was the same, in that we\u0027d be monkey patching one method vs another, but I may not have the full picture and if you tell me the other approach is more convoluted, this is good enough for me.","commit_id":"806cf71eb5af34359f8bc0c6b740752e53cf854f"},{"author":{"_account_id":11975,"name":"Slawek Kaplonski","email":"skaplons@redhat.com","username":"slaweq"},"change_message_id":"31d24b6ceef35f87de78f3a753b0c198974f2b51","unresolved":false,"context_lines":[{"line_number":211,"context_line":"    # https://bugs.launchpad.net/neutron/+bug/1687709 when more than one"},{"line_number":212,"context_line":"    # different ovs-agent with enabled trunk driver is running at a time it"},{"line_number":213,"context_line":"    # might lead to race contitions between them."},{"line_number":214,"context_line":"    # Because of that ovs_agent used for fullstack tests is monkeypatched and"},{"line_number":215,"context_line":"    # loads trunk driver only if trunk service plugin is enabled."},{"line_number":216,"context_line":"    # That makes restriction that only a single set of tests with trunk-enabled"},{"line_number":217,"context_line":"    # services will run at the same time."},{"line_number":218,"context_line":"    def test_trunk_lifecycle(self):"}],"source_content_type":"text/x-python","patch_set":2,"id":"3f4b6375_91c7f496","line":215,"range":{"start_line":214,"start_character":6,"end_line":215,"end_character":64},"in_reply_to":"3f4b6375_4e7e7daf","updated":"2017-11-06 10:02:37.000000000","message":"So do You think that I should propose new patch and change that?","commit_id":"806cf71eb5af34359f8bc0c6b740752e53cf854f"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"eb3ef6507837318af6947fd24f14500a2d9b3985","unresolved":false,"context_lines":[{"line_number":211,"context_line":"    # https://bugs.launchpad.net/neutron/+bug/1687709 when more than one"},{"line_number":212,"context_line":"    # different ovs-agent with enabled trunk driver is running at a time it"},{"line_number":213,"context_line":"    # might lead to race contitions between them."},{"line_number":214,"context_line":"    # Because of that ovs_agent used for fullstack tests is monkeypatched and"},{"line_number":215,"context_line":"    # loads trunk driver only if trunk service plugin is enabled."},{"line_number":216,"context_line":"    # That makes restriction that only a single set of tests with trunk-enabled"},{"line_number":217,"context_line":"    # services will run at the same time."},{"line_number":218,"context_line":"    def test_trunk_lifecycle(self):"}],"source_content_type":"text/x-python","patch_set":2,"id":"3f4b6375_4e7e7daf","line":215,"range":{"start_line":214,"start_character":6,"end_line":215,"end_character":64},"in_reply_to":"3f4b6375_77a34fb3","updated":"2017-11-06 09:38:04.000000000","message":"\u003e As we don\u0027t use nova to create the trunk bridge, server doesn\u0027t need to know about this change, right?\n\nCorrect, under the current circumstances.","commit_id":"806cf71eb5af34359f8bc0c6b740752e53cf854f"},{"author":{"_account_id":8655,"name":"Jakub Libosvar","email":"libosvar@redhat.com","username":"jlibosva"},"change_message_id":"7ac8c568273fa6c3cc3e0a1e67ee80d5bbb2a63b","unresolved":false,"context_lines":[{"line_number":211,"context_line":"    # https://bugs.launchpad.net/neutron/+bug/1687709 when more than one"},{"line_number":212,"context_line":"    # different ovs-agent with enabled trunk driver is running at a time it"},{"line_number":213,"context_line":"    # might lead to race contitions between them."},{"line_number":214,"context_line":"    # Because of that ovs_agent used for fullstack tests is monkeypatched and"},{"line_number":215,"context_line":"    # loads trunk driver only if trunk service plugin is enabled."},{"line_number":216,"context_line":"    # That makes restriction that only a single set of tests with trunk-enabled"},{"line_number":217,"context_line":"    # services will run at the same time."},{"line_number":218,"context_line":"    def test_trunk_lifecycle(self):"}],"source_content_type":"text/x-python","patch_set":2,"id":"3f4b6375_77a34fb3","line":215,"range":{"start_line":214,"start_character":6,"end_line":215,"end_character":64},"in_reply_to":"3f4b6375_caddce14","updated":"2017-11-06 05:11:29.000000000","message":"It should have been enough to just change https://github.com/openstack/neutron/blob/49d614895f44c44f9e1735210498facf1886c404/neutron/services/trunk/drivers/openvswitch/constants.py#L15 per ovs-agent process.\n\nAs we don\u0027t use nova to create the trunk bridge, server doesn\u0027t need to know about this change, right?","commit_id":"806cf71eb5af34359f8bc0c6b740752e53cf854f"},{"author":{"_account_id":11975,"name":"Slawek Kaplonski","email":"skaplons@redhat.com","username":"slaweq"},"change_message_id":"5b4607bb46f1f92fba7e174a0a24dada3d0b089e","unresolved":false,"context_lines":[{"line_number":211,"context_line":"    # https://bugs.launchpad.net/neutron/+bug/1687709 when more than one"},{"line_number":212,"context_line":"    # different ovs-agent with enabled trunk driver is running at a time it"},{"line_number":213,"context_line":"    # might lead to race contitions between them."},{"line_number":214,"context_line":"    # Because of that ovs_agent used for fullstack tests is monkeypatched and"},{"line_number":215,"context_line":"    # loads trunk driver only if trunk service plugin is enabled."},{"line_number":216,"context_line":"    # That makes restriction that only a single set of tests with trunk-enabled"},{"line_number":217,"context_line":"    # services will run at the same time."},{"line_number":218,"context_line":"    def test_trunk_lifecycle(self):"}],"source_content_type":"text/x-python","patch_set":2,"id":"3f4b6375_caddce14","line":215,"range":{"start_line":214,"start_character":6,"end_line":215,"end_character":64},"in_reply_to":"3f4b6375_e44d6472","updated":"2017-11-03 20:44:56.000000000","message":"I didn\u0027t try to do it with monkey patched methods to add test_prefix for bridges so I\u0027m not 100% sure how big it would be, but based on my previous \"experience\" with patch https://review.openstack.org/#/c/514586/ I think that it would require changing at least 2 or 3 different methods:\n\nis_trunk_bridge()\nTrunBridge.__init__()\nTrunkParentPort.dev_prefix()\nget_trunk_bridge_name()\n\nand use it in patched ovs_agent and also in test code.\nThis solution which I proposed here is smaller if You compare number of changed methods/functions - it\u0027s here only one function changed and in fact I have feeling that this change isn\u0027t big - it\u0027s \"just\" something like \"do nothing in one special case or execute standard production code\" and that\u0027s why I decided to try to do it this way.\nI\u0027m however not trunk expert so I might be wrong here. If I am wrong I can try to propose another patch with such approach and we will see which is better.","commit_id":"806cf71eb5af34359f8bc0c6b740752e53cf854f"},{"author":{"_account_id":11975,"name":"Slawek Kaplonski","email":"skaplons@redhat.com","username":"slaweq"},"change_message_id":"eaf6b1f2ccfa9c75050244e069cbebf48e24fe9b","unresolved":false,"context_lines":[{"line_number":211,"context_line":"    # https://bugs.launchpad.net/neutron/+bug/1687709 when more than one"},{"line_number":212,"context_line":"    # different ovs-agent with enabled trunk driver is running at a time it"},{"line_number":213,"context_line":"    # might lead to race contitions between them."},{"line_number":214,"context_line":"    # Because of that ovs_agent used for fullstack tests is monkeypatched and"},{"line_number":215,"context_line":"    # loads trunk driver only if trunk service plugin is enabled."},{"line_number":216,"context_line":"    # That makes restriction that only a single set of tests with trunk-enabled"},{"line_number":217,"context_line":"    # services will run at the same time."},{"line_number":218,"context_line":"    def test_trunk_lifecycle(self):"}],"source_content_type":"text/x-python","patch_set":2,"id":"3f4b6375_392285ba","line":215,"range":{"start_line":214,"start_character":6,"end_line":215,"end_character":64},"in_reply_to":"3f4b6375_f9d3cdff","updated":"2017-11-03 17:10:50.000000000","message":"I thought that it will be better as it\u0027s much smaller change and less code to monkey patch.\nIf I will monkey patch to generate bridge names with prefix I will have more code to change only for tests and also I will have to use this changed code in test runner.\nAs currently there is no other fullstack trunk tests planned (or I don\u0027t know about them) I decided to propose this solution.\nIf You think it would be better to use prefixes for bridge names I can change it of course.","commit_id":"806cf71eb5af34359f8bc0c6b740752e53cf854f"}]}
