)]}'
{"/COMMIT_MSG":[{"author":{"_account_id":6854,"name":"YAMAMOTO Takashi","email":"yamamoto@midokura.com","username":"yamamoto"},"change_message_id":"f226cb9f027c0c4d8a457cafb611f3e7cbde8f53","unresolved":false,"context_lines":[{"line_number":9,"context_line":"This commit partially reverts commit ced2e63cb8 in favor of"},{"line_number":10,"context_line":"update_device_{up,down}.  There\u0027s no limit on how many devices"},{"line_number":11,"context_line":"are updated per an update_device_list call and it can timeout."},{"line_number":12,"context_line":"Simply revert back to the old behavior."},{"line_number":13,"context_line":""},{"line_number":14,"context_line":"Change-Id: Ief2dfc0e87eb56d230d9d4033a773e1ca0d29c9c"},{"line_number":15,"context_line":"Closes-Bug: 1528895"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":1,"id":"5a5ae5dd_4bb4cc7e","line":12,"updated":"2016-02-08 05:37:01.000000000","message":"i tend to think this kind of problem needs a more general solution.\nsee https://bugs.launchpad.net/neutron/+bug/1430999","commit_id":"10b6066b05734d5a5731fd805762727702dd3e5f"},{"author":{"_account_id":9200,"name":"IWAMOTO Toshihiro","email":"iwamoto@valinux.co.jp","username":"toshii"},"change_message_id":"6cd809d167b08d5cc8a761c9baaf7b93796ad100","unresolved":false,"context_lines":[{"line_number":9,"context_line":"This commit partially reverts commit ced2e63cb8 in favor of"},{"line_number":10,"context_line":"update_device_{up,down}.  There\u0027s no limit on how many devices"},{"line_number":11,"context_line":"are updated per an update_device_list call and it can timeout."},{"line_number":12,"context_line":"Simply revert back to the old behavior."},{"line_number":13,"context_line":""},{"line_number":14,"context_line":"Change-Id: Ief2dfc0e87eb56d230d9d4033a773e1ca0d29c9c"},{"line_number":15,"context_line":"Closes-Bug: 1528895"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":1,"id":"5a5ae5dd_ebd497ec","line":12,"in_reply_to":"5a5ae5dd_4bb4cc7e","updated":"2016-02-08 09:26:37.000000000","message":"The bug you mentioned is related but different.\n\nAs update_device is more complicated than get_device_details, I don\u0027t think optimization by this bulk RPC call is large enough.  If we can optimize the server side, then it can be re-enabled on the agent side.","commit_id":"10b6066b05734d5a5731fd805762727702dd3e5f"},{"author":{"_account_id":6854,"name":"YAMAMOTO Takashi","email":"yamamoto@midokura.com","username":"yamamoto"},"change_message_id":"f226cb9f027c0c4d8a457cafb611f3e7cbde8f53","unresolved":false,"context_lines":[{"line_number":10,"context_line":"update_device_{up,down}.  There\u0027s no limit on how many devices"},{"line_number":11,"context_line":"are updated per an update_device_list call and it can timeout."},{"line_number":12,"context_line":"Simply revert back to the old behavior."},{"line_number":13,"context_line":""},{"line_number":14,"context_line":"Change-Id: Ief2dfc0e87eb56d230d9d4033a773e1ca0d29c9c"},{"line_number":15,"context_line":"Closes-Bug: 1528895"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":1,"id":"5a5ae5dd_ab5c607d","line":13,"updated":"2016-02-08 05:37:01.000000000","message":"mention blueprint restructure-l2-agent so that it\u0027s tracked","commit_id":"10b6066b05734d5a5731fd805762727702dd3e5f"},{"author":{"_account_id":9200,"name":"IWAMOTO Toshihiro","email":"iwamoto@valinux.co.jp","username":"toshii"},"change_message_id":"6cd809d167b08d5cc8a761c9baaf7b93796ad100","unresolved":false,"context_lines":[{"line_number":10,"context_line":"update_device_{up,down}.  There\u0027s no limit on how many devices"},{"line_number":11,"context_line":"are updated per an update_device_list call and it can timeout."},{"line_number":12,"context_line":"Simply revert back to the old behavior."},{"line_number":13,"context_line":""},{"line_number":14,"context_line":"Change-Id: Ief2dfc0e87eb56d230d9d4033a773e1ca0d29c9c"},{"line_number":15,"context_line":"Closes-Bug: 1528895"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":1,"id":"5a5ae5dd_da572356","line":13,"in_reply_to":"5a5ae5dd_ab5c607d","updated":"2016-02-08 09:26:37.000000000","message":"Done","commit_id":"10b6066b05734d5a5731fd805762727702dd3e5f"},{"author":{"_account_id":6854,"name":"YAMAMOTO Takashi","email":"yamamoto@midokura.com","username":"yamamoto"},"change_message_id":"b2303fe6ed0cb1055040bf82023afc4ef7ba1dd0","unresolved":false,"context_lines":[{"line_number":10,"context_line":"get_devices_details_list_and_failed_devices, have marginal performance"},{"line_number":11,"context_line":"benefit but don\u0027t scale as RPC timeouts are inherently constant and"},{"line_number":12,"context_line":"port density per server will increase in future.  Though the default"},{"line_number":13,"context_line":"limit of 100 is arbitrary, it preserves \u003e\u003d99% of performance benefit"},{"line_number":14,"context_line":"of the bulk RPC calls."},{"line_number":15,"context_line":""},{"line_number":16,"context_line":"Change-Id: Ief2dfc0e87eb56d230d9d4033a773e1ca0d29c9c"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":9,"id":"dab17558_26f1d995","line":13,"range":{"start_line":13,"start_character":40,"end_line":13,"end_character":45},"updated":"2016-05-27 05:37:12.000000000","message":"you just mean \"most\", or is there some calculation or benchmark behind this number?","commit_id":"cfc98105655ab4cc4a9019fde61ad9bd3be2d7c4"},{"author":{"_account_id":18289,"name":"Sreekumar S","email":"sreesiv@gmail.com","username":"sreesiv"},"change_message_id":"bb5a0cc7507352bd7400d29c5a96394b3ff2d598","unresolved":false,"context_lines":[{"line_number":10,"context_line":"get_devices_details_list_and_failed_devices, have marginal performance"},{"line_number":11,"context_line":"benefit but don\u0027t scale as RPC timeouts are inherently constant and"},{"line_number":12,"context_line":"port density per server will increase in future.  Though the default"},{"line_number":13,"context_line":"limit of 100 is arbitrary, it preserves \u003e\u003d99% of performance benefit"},{"line_number":14,"context_line":"of the bulk RPC calls."},{"line_number":15,"context_line":""},{"line_number":16,"context_line":"Change-Id: Ief2dfc0e87eb56d230d9d4033a773e1ca0d29c9c"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":9,"id":"9abb7d3a_9f4df19b","line":13,"range":{"start_line":13,"start_character":40,"end_line":13,"end_character":45},"in_reply_to":"9abb7d3a_a947ec49","updated":"2016-05-30 10:53:45.000000000","message":"I get your point, but I\u0027ve some comments here...\n\n\u003e\u003e I assumed constant overhead per rpc call.\nIMHO, you shouldn\u0027t assume that.\n\nAlso you haven\u0027t considered the operations on the server side, or the state of the server! It is when you consider these variance, that the issue with static config values arise!\n\nI believe this is a trade off situation. There is an advantage to do network calls in a group, but it\u0027s true only till a range. Beyond that threshold, the input size, and the server limit is not letting the client get a timely response. So we\u0027ve to try dynamically adjusting it for a fixed amount of time. If it\u0027s still not responding after a certain no of tries with adjusted size (specific to each call) and timeout, we\u0027ve to bail out!\n\nLike, if the size is 1500 and if the server can handle it, there is no point in doing it in 15 chunks. So grouping benefits here... But if you are grouping 15000, and cannot get a timely response, then grouping fails. There too, why do we need to do it 150 times?\n Try halving and check... and come down exponentially.\n\nThis way you get the benefit as described by your call saving formula \u003e\u003d (1 - 1/N). \u0027+\u0027 We get the benefit of grouping as well.\n\nWhat is your opinion?","commit_id":"cfc98105655ab4cc4a9019fde61ad9bd3be2d7c4"},{"author":{"_account_id":9200,"name":"IWAMOTO Toshihiro","email":"iwamoto@valinux.co.jp","username":"toshii"},"change_message_id":"806b25fd0415044c0e0a206d062990288ed9c0e8","unresolved":false,"context_lines":[{"line_number":10,"context_line":"get_devices_details_list_and_failed_devices, have marginal performance"},{"line_number":11,"context_line":"benefit but don\u0027t scale as RPC timeouts are inherently constant and"},{"line_number":12,"context_line":"port density per server will increase in future.  Though the default"},{"line_number":13,"context_line":"limit of 100 is arbitrary, it preserves \u003e\u003d99% of performance benefit"},{"line_number":14,"context_line":"of the bulk RPC calls."},{"line_number":15,"context_line":""},{"line_number":16,"context_line":"Change-Id: Ief2dfc0e87eb56d230d9d4033a773e1ca0d29c9c"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":9,"id":"9abb7d3a_a947ec49","line":13,"range":{"start_line":13,"start_character":40,"end_line":13,"end_character":45},"in_reply_to":"dab17558_26f1d995","updated":"2016-05-30 08:03:13.000000000","message":"I assumed constant overhead per rpc call.\n\nIf we chunk N devices per RPC call (of M total), the number of RPC calls is M/N.  If we have no limit, there\u0027ll be only one RPC call.\n\n(calls saved by N chunking) / (calls saved without limit) \u003d (M - M/N) / (M - 1) \u003e 1 - 1/N\n\nAssigning N\u003d100 gives 0.99.","commit_id":"cfc98105655ab4cc4a9019fde61ad9bd3be2d7c4"}],"neutron/plugins/ml2/drivers/openvswitch/agent/common/config.py":[{"author":{"_account_id":18289,"name":"Sreekumar S","email":"sreesiv@gmail.com","username":"sreesiv"},"change_message_id":"2504a60d198f97638d5495d6b47db63f644d1b2a","unresolved":false,"context_lines":[{"line_number":155,"context_line":"    cfg.StrOpt(\u0027agent_type\u0027, default\u003dn_const.AGENT_TYPE_OVS,"},{"line_number":156,"context_line":"               deprecated_for_removal\u003dTrue,"},{"line_number":157,"context_line":"               help\u003d_(\"Selects the Agent Type reported\")),"},{"line_number":158,"context_line":"    cfg.IntOpt(\u0027rpc_batch_limit\u0027, default\u003d100,"},{"line_number":159,"context_line":"               help\u003d_(\"Maximum argument size of RPC calls. Tune this only if \""},{"line_number":160,"context_line":"                      \"you see RPC timeouts\"))"},{"line_number":161,"context_line":"]"}],"source_content_type":"text/x-python","patch_set":9,"id":"9abb7d3a_50ac25a0","line":158,"range":{"start_line":158,"start_character":16,"end_line":158,"end_character":32},"updated":"2016-05-30 06:38:56.000000000","message":"I don\u0027t think having a config variable to judge the chunk size is the right solution here. There has been a very good debate on this https://review.openstack.org/#/c/163594/ and the core disagreed on fixing a predetermined chunk size.","commit_id":"cfc98105655ab4cc4a9019fde61ad9bd3be2d7c4"},{"author":{"_account_id":18289,"name":"Sreekumar S","email":"sreesiv@gmail.com","username":"sreesiv"},"change_message_id":"4e5d394335dadabb4ad0cdfb11a1558fac3d930a","unresolved":false,"context_lines":[{"line_number":155,"context_line":"    cfg.StrOpt(\u0027agent_type\u0027, default\u003dn_const.AGENT_TYPE_OVS,"},{"line_number":156,"context_line":"               deprecated_for_removal\u003dTrue,"},{"line_number":157,"context_line":"               help\u003d_(\"Selects the Agent Type reported\")),"},{"line_number":158,"context_line":"    cfg.IntOpt(\u0027rpc_batch_limit\u0027, default\u003d100,"},{"line_number":159,"context_line":"               help\u003d_(\"Maximum argument size of RPC calls. Tune this only if \""},{"line_number":160,"context_line":"                      \"you see RPC timeouts\"))"},{"line_number":161,"context_line":"]"}],"source_content_type":"text/x-python","patch_set":9,"id":"7aa08908_8cd683ef","line":158,"range":{"start_line":158,"start_character":16,"end_line":158,"end_character":32},"in_reply_to":"7aa08908_1bd8121d","updated":"2016-06-10 09:40:18.000000000","message":"\u003e\u003e needs 2x more time\nOnly when it times out! That\u0027s the point.\nKevin\u0027s patch is an initial solution to the blueprint you suggested earlier, it\u0027s mentioned in the RFE as well. It\u0027s so that the client does not overwhelm the server with more and more requests. So in a way batching is advantageous here.\n\n\u003e\u003e much slower than an agent without the batch RPC thing...\nNot exactly, IMHO we get this feeling, because the tests didn\u0027t emulate such a busy condition on the server. If the server is busy, then each of these small size chunk calls will take more time to come back, some may even fail with timeout!\n\nThat\u0027s why I tend to not agree on synthetic timing/call calculations.\n\n\u003e\u003e  It would make the whole thing rather slower.\nNot necessarily.\n\nNow imagine, that the server is moderately loaded, and still we are getting this issue consistently with port size beyond a certain threshold. I assume this is rare, and the bug here is not hitting such a scenario, but let\u0027s consider this case as well.\nIn this case the server is unable to enumerate all the details of all the ports requested in the timeout requested, and the client times out before the server had finished.\nNow there are two solutions to this rare case (rarity, I am assuming, because I don\u0027t have the resources to test and validate it :-))\n\n1) Provide a higher timeout value.\nIn this case Kevin\u0027s patch won\u0027t provide the adaptive timeout, but then the caller would have to adapt the timeout to the input size, or decide upon a large static timeout upfront (which is not good!)\n\n2) Config based chunk size\nHere too, we need to decide upon this chunk size upfront based on the server\u0027s threshold, which is again heavily based on the env (nw, hw, etc.). This is also not ideal IMHO.\n\n3) Adaptive size \u0026 timeout\n\nNow in this scenario we are assuming that the server is not timing out due to its load, but rather due to its inability to scale up to the input size. If that\u0027s the case, then it will fail each time it\u0027s called, and may end up taking more time each time (with batch RPC)! One solution would be to cache the last size that worked (adaptive case), and try with that the next time. But since I assume this is not the real cause of the bug, I felt there is no need to add it. But it\u0027s not a pain to include that as well along with adaptive chunk size.\n\nIf Kevin\u0027s BP is going to address this issue with push notifications/RPC call backs, then we can drop both the patches.\n\nBut core team has to comment on that.","commit_id":"cfc98105655ab4cc4a9019fde61ad9bd3be2d7c4"},{"author":{"_account_id":18289,"name":"Sreekumar S","email":"sreesiv@gmail.com","username":"sreesiv"},"change_message_id":"636caf796a6d1d0e22a8cce92f45b86e5a1eeb28","unresolved":false,"context_lines":[{"line_number":155,"context_line":"    cfg.StrOpt(\u0027agent_type\u0027, default\u003dn_const.AGENT_TYPE_OVS,"},{"line_number":156,"context_line":"               deprecated_for_removal\u003dTrue,"},{"line_number":157,"context_line":"               help\u003d_(\"Selects the Agent Type reported\")),"},{"line_number":158,"context_line":"    cfg.IntOpt(\u0027rpc_batch_limit\u0027, default\u003d100,"},{"line_number":159,"context_line":"               help\u003d_(\"Maximum argument size of RPC calls. Tune this only if \""},{"line_number":160,"context_line":"                      \"you see RPC timeouts\"))"},{"line_number":161,"context_line":"]"}],"source_content_type":"text/x-python","patch_set":9,"id":"7aa08908_eb7a1487","line":158,"range":{"start_line":158,"start_character":16,"end_line":158,"end_character":32},"in_reply_to":"7aa08908_e4ab6c73","updated":"2016-06-08 10:37:43.000000000","message":"I think, we need to retry the operation with adaptive chunk size, even with adaptive timeout (Kevin\u0027s patch - https://review.openstack.org/#/c/280595).","commit_id":"cfc98105655ab4cc4a9019fde61ad9bd3be2d7c4"},{"author":{"_account_id":9200,"name":"IWAMOTO Toshihiro","email":"iwamoto@valinux.co.jp","username":"toshii"},"change_message_id":"0b8ab775d550750fa28671811e9a1af83d297c42","unresolved":false,"context_lines":[{"line_number":155,"context_line":"    cfg.StrOpt(\u0027agent_type\u0027, default\u003dn_const.AGENT_TYPE_OVS,"},{"line_number":156,"context_line":"               deprecated_for_removal\u003dTrue,"},{"line_number":157,"context_line":"               help\u003d_(\"Selects the Agent Type reported\")),"},{"line_number":158,"context_line":"    cfg.IntOpt(\u0027rpc_batch_limit\u0027, default\u003d100,"},{"line_number":159,"context_line":"               help\u003d_(\"Maximum argument size of RPC calls. Tune this only if \""},{"line_number":160,"context_line":"                      \"you see RPC timeouts\"))"},{"line_number":161,"context_line":"]"}],"source_content_type":"text/x-python","patch_set":9,"id":"7aa08908_1bd8121d","line":158,"range":{"start_line":158,"start_character":16,"end_line":158,"end_character":32},"in_reply_to":"7aa08908_e4ab6c73","updated":"2016-06-10 08:02:18.000000000","message":"With the 280595, an agent eventually succeeds in finishing the second rpc_loop without errors, only after an RPC timeout and an aborted rpc_loop.  This means an agent needs 2x more time until a successful sync, which is much slower than an agent without the batch RPC thing.\n\nGiven the amount of improvement by batch RPC, we cannot afford timeout-and-retry.  It would make the whole thing rather slower.","commit_id":"cfc98105655ab4cc4a9019fde61ad9bd3be2d7c4"},{"author":{"_account_id":9200,"name":"IWAMOTO Toshihiro","email":"iwamoto@valinux.co.jp","username":"toshii"},"change_message_id":"8d2920613bc1fb9694a8a17a968a63a82b814fd9","unresolved":false,"context_lines":[{"line_number":155,"context_line":"    cfg.StrOpt(\u0027agent_type\u0027, default\u003dn_const.AGENT_TYPE_OVS,"},{"line_number":156,"context_line":"               deprecated_for_removal\u003dTrue,"},{"line_number":157,"context_line":"               help\u003d_(\"Selects the Agent Type reported\")),"},{"line_number":158,"context_line":"    cfg.IntOpt(\u0027rpc_batch_limit\u0027, default\u003d100,"},{"line_number":159,"context_line":"               help\u003d_(\"Maximum argument size of RPC calls. Tune this only if \""},{"line_number":160,"context_line":"                      \"you see RPC timeouts\"))"},{"line_number":161,"context_line":"]"}],"source_content_type":"text/x-python","patch_set":9,"id":"7aa08908_de45186f","line":158,"range":{"start_line":158,"start_character":16,"end_line":158,"end_character":32},"in_reply_to":"7aa08908_eb7a1487","updated":"2016-06-09 09:23:10.000000000","message":"I agree rpc_batch_limit option is not ideal but this is not where fine tuning is needed.  We already have adaptive timeout. I don\u0027t think extra complexity is justifiable.","commit_id":"cfc98105655ab4cc4a9019fde61ad9bd3be2d7c4"},{"author":{"_account_id":9200,"name":"IWAMOTO Toshihiro","email":"iwamoto@valinux.co.jp","username":"toshii"},"change_message_id":"806b25fd0415044c0e0a206d062990288ed9c0e8","unresolved":false,"context_lines":[{"line_number":155,"context_line":"    cfg.StrOpt(\u0027agent_type\u0027, default\u003dn_const.AGENT_TYPE_OVS,"},{"line_number":156,"context_line":"               deprecated_for_removal\u003dTrue,"},{"line_number":157,"context_line":"               help\u003d_(\"Selects the Agent Type reported\")),"},{"line_number":158,"context_line":"    cfg.IntOpt(\u0027rpc_batch_limit\u0027, default\u003d100,"},{"line_number":159,"context_line":"               help\u003d_(\"Maximum argument size of RPC calls. Tune this only if \""},{"line_number":160,"context_line":"                      \"you see RPC timeouts\"))"},{"line_number":161,"context_line":"]"}],"source_content_type":"text/x-python","patch_set":9,"id":"9abb7d3a_a69b9b71","line":158,"range":{"start_line":158,"start_character":16,"end_line":158,"end_character":32},"in_reply_to":"9abb7d3a_50ac25a0","updated":"2016-05-30 08:03:13.000000000","message":"I don\u0027t agree on the argument made on that patch.\n\nOn an unloaded not-very-old-ish 3GHz CPU machine, the limit is around 250 or 500 devices. (See https://bugs.launchpad.net/neutron/+bug/1528895/comments/9 )\n\nMaybe we need a larger safety factor, but IMO there\u0027s little point in chunking as much as devices in an RPC call.","commit_id":"cfc98105655ab4cc4a9019fde61ad9bd3be2d7c4"},{"author":{"_account_id":18289,"name":"Sreekumar S","email":"sreesiv@gmail.com","username":"sreesiv"},"change_message_id":"bb5a0cc7507352bd7400d29c5a96394b3ff2d598","unresolved":false,"context_lines":[{"line_number":155,"context_line":"    cfg.StrOpt(\u0027agent_type\u0027, default\u003dn_const.AGENT_TYPE_OVS,"},{"line_number":156,"context_line":"               deprecated_for_removal\u003dTrue,"},{"line_number":157,"context_line":"               help\u003d_(\"Selects the Agent Type reported\")),"},{"line_number":158,"context_line":"    cfg.IntOpt(\u0027rpc_batch_limit\u0027, default\u003d100,"},{"line_number":159,"context_line":"               help\u003d_(\"Maximum argument size of RPC calls. Tune this only if \""},{"line_number":160,"context_line":"                      \"you see RPC timeouts\"))"},{"line_number":161,"context_line":"]"}],"source_content_type":"text/x-python","patch_set":9,"id":"9abb7d3a_e2a8e6dd","line":158,"range":{"start_line":158,"start_character":16,"end_line":158,"end_character":32},"in_reply_to":"9abb7d3a_a69b9b71","updated":"2016-05-30 10:53:45.000000000","message":"OK. Your prediction came true. \nTime for other experts to chime in. :-)","commit_id":"cfc98105655ab4cc4a9019fde61ad9bd3be2d7c4"},{"author":{"_account_id":6788,"name":"Rossella Sblendido","email":"rsblendido@suse.com","username":"rossella-o"},"change_message_id":"ab8cdc30cd07e5dd3fc4108b03490b103db4dc1a","unresolved":false,"context_lines":[{"line_number":155,"context_line":"    cfg.StrOpt(\u0027agent_type\u0027, default\u003dn_const.AGENT_TYPE_OVS,"},{"line_number":156,"context_line":"               deprecated_for_removal\u003dTrue,"},{"line_number":157,"context_line":"               help\u003d_(\"Selects the Agent Type reported\")),"},{"line_number":158,"context_line":"    cfg.IntOpt(\u0027rpc_batch_limit\u0027, default\u003d100,"},{"line_number":159,"context_line":"               help\u003d_(\"Maximum argument size of RPC calls. Tune this only if \""},{"line_number":160,"context_line":"                      \"you see RPC timeouts\"))"},{"line_number":161,"context_line":"]"}],"source_content_type":"text/x-python","patch_set":9,"id":"7aa08908_e4ab6c73","line":158,"range":{"start_line":158,"start_character":16,"end_line":158,"end_character":32},"in_reply_to":"9abb7d3a_e2a8e6dd","updated":"2016-06-08 08:41:51.000000000","message":"a new config option is not ideal. Iwamoto did you test this after the merge of https://review.openstack.org/#/c/280595 ? Is the bug still present? I hope not! If yes then we could try to have an adaptive chunk size. When we get a timeout we set the chunk size equal to the number of devices/2 ...","commit_id":"cfc98105655ab4cc4a9019fde61ad9bd3be2d7c4"}],"neutron/plugins/ml2/drivers/openvswitch/agent/ovs_neutron_agent.py":[{"author":{"_account_id":18289,"name":"Sreekumar S","email":"sreesiv@gmail.com","username":"sreesiv"},"change_message_id":"2504a60d198f97638d5495d6b47db63f644d1b2a","unresolved":false,"context_lines":[{"line_number":828,"context_line":"                    \"Port\", port.port_name, \"other_config\", other_config)"},{"line_number":829,"context_line":""},{"line_number":830,"context_line":"    def _bind_devices_subr(self, devices_up, devices_down):"},{"line_number":831,"context_line":"        devices_set \u003d self.plugin_rpc.update_device_list("},{"line_number":832,"context_line":"            self.context, devices_up, devices_down, self.agent_id,"},{"line_number":833,"context_line":"            self.conf.host)"},{"line_number":834,"context_line":"        failed_devices \u003d (devices_set.get(\u0027failed_devices_up\u0027) +"},{"line_number":835,"context_line":"            devices_set.get(\u0027failed_devices_down\u0027))"},{"line_number":836,"context_line":"        if failed_devices:"}],"source_content_type":"text/x-python","patch_set":9,"id":"9abb7d3a_b0d751ec","line":833,"range":{"start_line":831,"start_character":0,"end_line":833,"end_character":27},"updated":"2016-05-30 06:38:56.000000000","message":"There is no retry attempted taking advantage of the latest RPC exponential back-off implementation!","commit_id":"cfc98105655ab4cc4a9019fde61ad9bd3be2d7c4"},{"author":{"_account_id":9200,"name":"IWAMOTO Toshihiro","email":"iwamoto@valinux.co.jp","username":"toshii"},"change_message_id":"806b25fd0415044c0e0a206d062990288ed9c0e8","unresolved":false,"context_lines":[{"line_number":828,"context_line":"                    \"Port\", port.port_name, \"other_config\", other_config)"},{"line_number":829,"context_line":""},{"line_number":830,"context_line":"    def _bind_devices_subr(self, devices_up, devices_down):"},{"line_number":831,"context_line":"        devices_set \u003d self.plugin_rpc.update_device_list("},{"line_number":832,"context_line":"            self.context, devices_up, devices_down, self.agent_id,"},{"line_number":833,"context_line":"            self.conf.host)"},{"line_number":834,"context_line":"        failed_devices \u003d (devices_set.get(\u0027failed_devices_up\u0027) +"},{"line_number":835,"context_line":"            devices_set.get(\u0027failed_devices_down\u0027))"},{"line_number":836,"context_line":"        if failed_devices:"}],"source_content_type":"text/x-python","patch_set":9,"id":"9abb7d3a_e62e43f9","line":833,"range":{"start_line":831,"start_character":0,"end_line":833,"end_character":27},"in_reply_to":"9abb7d3a_b0d751ec","updated":"2016-05-30 08:03:13.000000000","message":"Doesn\u0027t PluginApi do that automatically?","commit_id":"cfc98105655ab4cc4a9019fde61ad9bd3be2d7c4"},{"author":{"_account_id":18289,"name":"Sreekumar S","email":"sreesiv@gmail.com","username":"sreesiv"},"change_message_id":"2504a60d198f97638d5495d6b47db63f644d1b2a","unresolved":false,"context_lines":[{"line_number":884,"context_line":"            else:"},{"line_number":885,"context_line":"                LOG.debug(\"Setting status for %s to DOWN\", device)"},{"line_number":886,"context_line":"                devices_down.append(device)"},{"line_number":887,"context_line":"            if (len(devices_up) + len(devices_down) \u003e\u003d"},{"line_number":888,"context_line":"                self.conf.AGENT.rpc_batch_limit):"},{"line_number":889,"context_line":"                failed_devices.extend("},{"line_number":890,"context_line":"                    self._bind_devices_subr(devices_up, devices_down))"},{"line_number":891,"context_line":"                devices_up \u003d []"}],"source_content_type":"text/x-python","patch_set":9,"id":"9abb7d3a_1013bdb9","line":888,"range":{"start_line":887,"start_character":48,"end_line":888,"end_character":16},"updated":"2016-05-30 06:38:56.000000000","message":"Predetermined chunk size.","commit_id":"cfc98105655ab4cc4a9019fde61ad9bd3be2d7c4"},{"author":{"_account_id":18289,"name":"Sreekumar S","email":"sreesiv@gmail.com","username":"sreesiv"},"change_message_id":"2504a60d198f97638d5495d6b47db63f644d1b2a","unresolved":false,"context_lines":[{"line_number":1494,"context_line":"        devices \u003d list(devices)"},{"line_number":1495,"context_line":"        for i in range(0, len(devices), self.conf.AGENT.rpc_batch_limit):"},{"line_number":1496,"context_line":"            ret \u003d ("},{"line_number":1497,"context_line":"                self.plugin_rpc.get_devices_details_list_and_failed_devices("},{"line_number":1498,"context_line":"                    self.context,"},{"line_number":1499,"context_line":"                    devices[i:(i + self.conf.AGENT.rpc_batch_limit)],"},{"line_number":1500,"context_line":"                    self.agent_id,"},{"line_number":1501,"context_line":"                    self.conf.host))"},{"line_number":1502,"context_line":"            devices_details.extend(ret.get(\u0027devices\u0027))"},{"line_number":1503,"context_line":"            failed_devices.extend(ret.get(\u0027failed_devices\u0027))"},{"line_number":1504,"context_line":""}],"source_content_type":"text/x-python","patch_set":9,"id":"9abb7d3a_102c5d01","line":1501,"range":{"start_line":1497,"start_character":0,"end_line":1501,"end_character":36},"updated":"2016-05-30 06:38:56.000000000","message":"Same here, no retry attempted + predetermined chunk size.","commit_id":"cfc98105655ab4cc4a9019fde61ad9bd3be2d7c4"}]}
