)]}'
{"/COMMIT_MSG":[{"author":{"_account_id":5948,"name":"Oleg Bondarev","email":"obondarev@mirantis.com","username":"obondarev"},"change_message_id":"030325b0b5b50b7373105c664bfdf390e124b746","unresolved":true,"context_lines":[{"line_number":10,"context_line":"to deal with large sets always equals 100."},{"line_number":11,"context_line":"Number of resources for neutron to divide"},{"line_number":12,"context_line":"the large RPC call data sets always equals 20."},{"line_number":13,"context_line":"In \"big\" networks even these numbers can be"},{"line_number":14,"context_line":"still huge and cause service timeouts."},{"line_number":15,"context_line":"Now we can decrease these numbers in config."},{"line_number":16,"context_line":""}],"source_content_type":"text/x-gerrit-commit-message","patch_set":3,"id":"30c6dfce_9866b63c","line":13,"range":{"start_line":13,"start_character":0,"end_line":13,"end_character":17},"updated":"2021-07-28 05:21:54.000000000","message":"How could \"big\" network cause timeout? timeout may be caused by chunk size, no matter if it\u0027s one big network or many networks of size 100. I guess the reason for the bug is high load on physical node, or lack of hw resources. Please clarify in the bug.\n\nAlso agent step for OVS flows and RPC step are very different things. Are you hitting both types of timeouts?\n\nPlease share which HW resources are you running OVS agent and Neutron server on.","commit_id":"00c2199b426e9123aafb5c7adad6f4a0c47e91bd"},{"author":{"_account_id":32927,"name":"mitya-eremeev-2","display_name":"Mitya Eremeev","email":"mitossvyaz@mail.ru","username":"mitya-eremeev-2"},"change_message_id":"812838027ea8519f78fd0bba8fe03acd71987d0e","unresolved":true,"context_lines":[{"line_number":10,"context_line":"to deal with large sets always equals 100."},{"line_number":11,"context_line":"Number of resources for neutron to divide"},{"line_number":12,"context_line":"the large RPC call data sets always equals 20."},{"line_number":13,"context_line":"In \"big\" networks even these numbers can be"},{"line_number":14,"context_line":"still huge and cause service timeouts."},{"line_number":15,"context_line":"Now we can decrease these numbers in config."},{"line_number":16,"context_line":""}],"source_content_type":"text/x-gerrit-commit-message","patch_set":3,"id":"eed76131_2fbfe0be","line":13,"range":{"start_line":13,"start_character":0,"end_line":13,"end_character":17},"in_reply_to":"30c6dfce_9866b63c","updated":"2021-07-28 09:59:03.000000000","message":"Oleg, thank you for feedback.\nI updated bug description.\nIf more information required please ask.","commit_id":"00c2199b426e9123aafb5c7adad6f4a0c47e91bd"},{"author":{"_account_id":32927,"name":"mitya-eremeev-2","display_name":"Mitya Eremeev","email":"mitossvyaz@mail.ru","username":"mitya-eremeev-2"},"change_message_id":"879de692f7246d7626692dc02e30b26a3bad7374","unresolved":true,"context_lines":[{"line_number":10,"context_line":"to deal with large sets always equals 100."},{"line_number":11,"context_line":"Number of resources for neutron to divide"},{"line_number":12,"context_line":"the large RPC call data sets always equals 20."},{"line_number":13,"context_line":"In \"big\" networks even these numbers can be"},{"line_number":14,"context_line":"still huge and cause service timeouts."},{"line_number":15,"context_line":"Now we can decrease these numbers in config."},{"line_number":16,"context_line":""}],"source_content_type":"text/x-gerrit-commit-message","patch_set":3,"id":"1e712110_b7e67c17","line":13,"range":{"start_line":13,"start_character":0,"end_line":13,"end_character":17},"in_reply_to":"71e5a455_485fa12b","updated":"2021-07-28 12:11:39.000000000","message":"Yes, new 2 config options help to reduce timeouts, but performance suffers which is expected behavior.\nImportant to notice that patch does not introduce neutron performance issues.\nDefault values of new 2 options have the same values as hard-coded constants which were used before.\nNew options are additional tuning if user is not happy with default behavior.","commit_id":"00c2199b426e9123aafb5c7adad6f4a0c47e91bd"},{"author":{"_account_id":5948,"name":"Oleg Bondarev","email":"obondarev@mirantis.com","username":"obondarev"},"change_message_id":"6d0b9c83a3d73fdb4aecf7ce88b7df1a1d7e8079","unresolved":true,"context_lines":[{"line_number":10,"context_line":"to deal with large sets always equals 100."},{"line_number":11,"context_line":"Number of resources for neutron to divide"},{"line_number":12,"context_line":"the large RPC call data sets always equals 20."},{"line_number":13,"context_line":"In \"big\" networks even these numbers can be"},{"line_number":14,"context_line":"still huge and cause service timeouts."},{"line_number":15,"context_line":"Now we can decrease these numbers in config."},{"line_number":16,"context_line":""}],"source_content_type":"text/x-gerrit-commit-message","patch_set":3,"id":"71e5a455_485fa12b","line":13,"range":{"start_line":13,"start_character":0,"end_line":13,"end_character":17},"in_reply_to":"eed76131_2fbfe0be","updated":"2021-07-28 10:51:22.000000000","message":"Ok, it\u0027s more clear now, thanks.\nHowever the two timeouts this patch is touching seem mostly affect agent restart case - when agent needs to reprocess all hosted ports. From bug description it looks the issue is with single port processing - just the amount of info needed is very big (especially if the port is first port on the node from this network) given the size of the network.\n\nDid you managed to overcome the issue described by adjusting these 2 timeouts?","commit_id":"00c2199b426e9123aafb5c7adad6f4a0c47e91bd"},{"author":{"_account_id":9531,"name":"liuyulong","display_name":"LIU Yulong","email":"i@liuyulong.me","username":"LIU-Yulong"},"change_message_id":"a1376d40dc0872b83ce6eb9e84b8259239b3d130","unresolved":true,"context_lines":[{"line_number":11,"context_line":"Number of resources for neutron to divide"},{"line_number":12,"context_line":"the large RPC call data sets always equals 20."},{"line_number":13,"context_line":"In \"big\" networks even these numbers can be"},{"line_number":14,"context_line":"still huge and cause service timeouts."},{"line_number":15,"context_line":"Now we can decrease these numbers in config."},{"line_number":16,"context_line":""},{"line_number":17,"context_line":"Closes-Bug: 1938202"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":3,"id":"e085bf1a_55cba27f","line":14,"range":{"start_line":14,"start_character":15,"end_line":14,"end_character":38},"updated":"2021-07-29 11:55:13.000000000","message":"Form the bug description, this seems to be the RPC timeout. So please increase the RPC time out option to overcome large set data actions.","commit_id":"00c2199b426e9123aafb5c7adad6f4a0c47e91bd"},{"author":{"_account_id":32927,"name":"mitya-eremeev-2","display_name":"Mitya Eremeev","email":"mitossvyaz@mail.ru","username":"mitya-eremeev-2"},"change_message_id":"f4ec080ff7bff5b0ba26995cc4e2d2b5c7a7e4dc","unresolved":true,"context_lines":[{"line_number":11,"context_line":"Number of resources for neutron to divide"},{"line_number":12,"context_line":"the large RPC call data sets always equals 20."},{"line_number":13,"context_line":"In \"big\" networks even these numbers can be"},{"line_number":14,"context_line":"still huge and cause service timeouts."},{"line_number":15,"context_line":"Now we can decrease these numbers in config."},{"line_number":16,"context_line":""},{"line_number":17,"context_line":"Closes-Bug: 1938202"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":3,"id":"2c0bfe19_0011eeb7","line":14,"range":{"start_line":14,"start_character":15,"end_line":14,"end_character":38},"in_reply_to":"11251e5d_1aef1ace","updated":"2021-07-29 12:38:35.000000000","message":"If it\u0027s about rpc_response_max_timeout, then I don\u0027t see where the config option is used in code.","commit_id":"00c2199b426e9123aafb5c7adad6f4a0c47e91bd"},{"author":{"_account_id":32927,"name":"mitya-eremeev-2","display_name":"Mitya Eremeev","email":"mitossvyaz@mail.ru","username":"mitya-eremeev-2"},"change_message_id":"b1c17cc5ff4a0e0d9caaf4a163669933cab943ff","unresolved":true,"context_lines":[{"line_number":11,"context_line":"Number of resources for neutron to divide"},{"line_number":12,"context_line":"the large RPC call data sets always equals 20."},{"line_number":13,"context_line":"In \"big\" networks even these numbers can be"},{"line_number":14,"context_line":"still huge and cause service timeouts."},{"line_number":15,"context_line":"Now we can decrease these numbers in config."},{"line_number":16,"context_line":""},{"line_number":17,"context_line":"Closes-Bug: 1938202"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":3,"id":"65bc3dd1_99a2fc25","line":14,"range":{"start_line":14,"start_character":15,"end_line":14,"end_character":38},"in_reply_to":"29e00791_74303914","updated":"2021-07-30 10:03:40.000000000","message":"Sorry for my poor explanation.\nWe have increased values of both config options: rpc_response_max_timeout and rpc_response_timeout.\nI agree it\u0027s not the smartest idea to touch all possible options)))\nI found that rpc_response_timeout is used in code, but did not find rpc_response_max_timeout in code.\nThat\u0027s why I asked the question.\nThank you for clarification.","commit_id":"00c2199b426e9123aafb5c7adad6f4a0c47e91bd"},{"author":{"_account_id":9531,"name":"liuyulong","display_name":"LIU Yulong","email":"i@liuyulong.me","username":"LIU-Yulong"},"change_message_id":"0bbca7c48b1a1fecf192116b7636797d170cf031","unresolved":true,"context_lines":[{"line_number":11,"context_line":"Number of resources for neutron to divide"},{"line_number":12,"context_line":"the large RPC call data sets always equals 20."},{"line_number":13,"context_line":"In \"big\" networks even these numbers can be"},{"line_number":14,"context_line":"still huge and cause service timeouts."},{"line_number":15,"context_line":"Now we can decrease these numbers in config."},{"line_number":16,"context_line":""},{"line_number":17,"context_line":"Closes-Bug: 1938202"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":3,"id":"29e00791_74303914","line":14,"range":{"start_line":14,"start_character":15,"end_line":14,"end_character":38},"in_reply_to":"2c0bfe19_0011eeb7","updated":"2021-07-30 01:45:31.000000000","message":"IMO, you set the wrong option. It should be \"rpc_response_timeout\" which is forked from oslo.messaging. It\u0027s default value is 60, it is smaller than your 120s in the bug description. Things look clear now. 😊\n\n[1] https://github.com/openstack/oslo.messaging/blob/master/oslo_messaging/rpc/client.py#L39","commit_id":"00c2199b426e9123aafb5c7adad6f4a0c47e91bd"},{"author":{"_account_id":32927,"name":"mitya-eremeev-2","display_name":"Mitya Eremeev","email":"mitossvyaz@mail.ru","username":"mitya-eremeev-2"},"change_message_id":"219d32142616cb37555c09c37dbda3a612050347","unresolved":true,"context_lines":[{"line_number":11,"context_line":"Number of resources for neutron to divide"},{"line_number":12,"context_line":"the large RPC call data sets always equals 20."},{"line_number":13,"context_line":"In \"big\" networks even these numbers can be"},{"line_number":14,"context_line":"still huge and cause service timeouts."},{"line_number":15,"context_line":"Now we can decrease these numbers in config."},{"line_number":16,"context_line":""},{"line_number":17,"context_line":"Closes-Bug: 1938202"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":3,"id":"11251e5d_1aef1ace","line":14,"range":{"start_line":14,"start_character":15,"end_line":14,"end_character":38},"in_reply_to":"e085bf1a_55cba27f","updated":"2021-07-29 11:58:43.000000000","message":"RPC timeout was already increased - it doesn\u0027t help any more.","commit_id":"00c2199b426e9123aafb5c7adad6f4a0c47e91bd"}],"neutron/agent/common/ovs_lib.py":[{"author":{"_account_id":9531,"name":"liuyulong","display_name":"LIU Yulong","email":"i@liuyulong.me","username":"LIU-Yulong"},"change_message_id":"0bbca7c48b1a1fecf192116b7636797d170cf031","unresolved":true,"context_lines":[{"line_number":467,"context_line":"            if use_bundle:"},{"line_number":468,"context_line":"                extra_param.append(\u0027--bundle\u0027)"},{"line_number":469,"context_line":""},{"line_number":470,"context_line":"            step \u003d cfg.CONF.agent_resources_processing_step"},{"line_number":471,"context_line":"            for i in range(0, len(flow_strs), step):"},{"line_number":472,"context_line":"                self.run_ofctl(\u0027%s-flows\u0027 % action, extra_param + [\u0027-\u0027],"},{"line_number":473,"context_line":"                               \u0027\\n\u0027.join(flow_strs[i:i + step]))"}],"source_content_type":"text/x-python","patch_set":3,"id":"075f6a3a_462f5a33","line":470,"range":{"start_line":470,"start_character":12,"end_line":470,"end_character":59},"updated":"2021-07-30 01:45:31.000000000","message":"Let\u0027s make this work automatically in patch https://review.opendev.org/c/openstack/neutron/+/765072.","commit_id":"00c2199b426e9123aafb5c7adad6f4a0c47e91bd"},{"author":{"_account_id":9531,"name":"liuyulong","display_name":"LIU Yulong","email":"i@liuyulong.me","username":"LIU-Yulong"},"change_message_id":"4bfaf9a5dd71e08e43a2175aecf3bc2bbbef687c","unresolved":true,"context_lines":[{"line_number":466,"context_line":"            LOG.debug(\"Processing %d OpenFlow rules.\", len(flow_strs))"},{"line_number":467,"context_line":"            if use_bundle:"},{"line_number":468,"context_line":"                extra_param.append(\u0027--bundle\u0027)"},{"line_number":469,"context_line":"            step \u003d cfg.CONF.OVS.ovs_flow_string_processing_step"},{"line_number":470,"context_line":"            for i in range(0, len(flow_strs), step):"},{"line_number":471,"context_line":"                self.run_ofctl(\u0027%s-flows\u0027 % action, extra_param + [\u0027-\u0027],"},{"line_number":472,"context_line":"                               \u0027\\n\u0027.join(flow_strs[i:i + step]))"}],"source_content_type":"text/x-python","patch_set":10,"id":"ffa97f61_3a08f17c","line":469,"range":{"start_line":469,"start_character":12,"end_line":469,"end_character":63},"updated":"2021-08-13 15:40:07.000000000","message":"I\u0027d perfer not touch this. Please remember the patch here: https://review.opendev.org/c/openstack/neutron/+/765072","commit_id":"9b530b0040624203862d1376a73e837663a4aeb5"}],"neutron/agent/rpc.py":[{"author":{"_account_id":33145,"name":"Kevin Li","email":"liwenhai3@huawei.com","username":"liwenhai"},"change_message_id":"3c11a8e5ae9f323f0c9b2ea94d433b5386b0561e","unresolved":true,"context_lines":[{"line_number":182,"context_line":"        ret_devices_down \u003d []"},{"line_number":183,"context_line":"        failed_devices_down \u003d []"},{"line_number":184,"context_line":""},{"line_number":185,"context_line":"        step \u003d n_const.RPC_RES_PROCESSING_STEP"},{"line_number":186,"context_line":"        devices_up \u003d list(devices_up)"},{"line_number":187,"context_line":"        devices_down \u003d list(devices_down)"},{"line_number":188,"context_line":"        for i in range(0, max(len(devices_up), len(devices_down)), step):"}],"source_content_type":"text/x-python","patch_set":9,"id":"8c1bea92_a11a0c79","side":"PARENT","line":185,"range":{"start_line":185,"start_character":15,"end_line":185,"end_character":46},"updated":"2021-08-12 07:48:32.000000000","message":"RPC_RES_PROCESSING_STEP \u003d 20 is not useful any more, it need to be removed from neutron/common/_constants.py +62","commit_id":"476bf747832694faee92a28fc86a04209da7eca9"},{"author":{"_account_id":32927,"name":"mitya-eremeev-2","display_name":"Mitya Eremeev","email":"mitossvyaz@mail.ru","username":"mitya-eremeev-2"},"change_message_id":"0671ddd29eb56729c1ba456db490bcf77f557491","unresolved":true,"context_lines":[{"line_number":182,"context_line":"        ret_devices_down \u003d []"},{"line_number":183,"context_line":"        failed_devices_down \u003d []"},{"line_number":184,"context_line":""},{"line_number":185,"context_line":"        step \u003d n_const.RPC_RES_PROCESSING_STEP"},{"line_number":186,"context_line":"        devices_up \u003d list(devices_up)"},{"line_number":187,"context_line":"        devices_down \u003d list(devices_down)"},{"line_number":188,"context_line":"        for i in range(0, max(len(devices_up), len(devices_down)), step):"}],"source_content_type":"text/x-python","patch_set":9,"id":"650e1560_3328a8ae","side":"PARENT","line":185,"range":{"start_line":185,"start_character":15,"end_line":185,"end_character":46},"in_reply_to":"8c1bea92_a11a0c79","updated":"2021-08-12 14:40:53.000000000","message":"done","commit_id":"476bf747832694faee92a28fc86a04209da7eca9"}],"neutron/agent/securitygroups_rpc.py":[{"author":{"_account_id":9531,"name":"liuyulong","display_name":"LIU Yulong","email":"i@liuyulong.me","username":"LIU-Yulong"},"change_message_id":"0bbca7c48b1a1fecf192116b7636797d170cf031","unresolved":true,"context_lines":[{"line_number":154,"context_line":""},{"line_number":155,"context_line":"    @_port_filter_lock"},{"line_number":156,"context_line":"    def _apply_port_filter(self, device_ids, update_filter\u003dFalse):"},{"line_number":157,"context_line":"        step \u003d cfg.CONF.agent_resources_processing_step"},{"line_number":158,"context_line":"        devices \u003d {}"},{"line_number":159,"context_line":"        security_groups \u003d {}"},{"line_number":160,"context_line":"        security_group_member_ips \u003d {}"}],"source_content_type":"text/x-python","patch_set":3,"id":"0958f2a0_487afda0","line":157,"range":{"start_line":157,"start_character":15,"end_line":157,"end_character":55},"updated":"2021-07-30 01:45:31.000000000","message":"Increasing this step size can cause RPC timeout when you have large sets resources and restart many agents in a short time.\n\nReducing this step size can make your agent restart time too long to process new resources.\n\nSo, IMO, 20 is a bearable value from our local long time experiences.","commit_id":"00c2199b426e9123aafb5c7adad6f4a0c47e91bd"},{"author":{"_account_id":32927,"name":"mitya-eremeev-2","display_name":"Mitya Eremeev","email":"mitossvyaz@mail.ru","username":"mitya-eremeev-2"},"change_message_id":"b1c17cc5ff4a0e0d9caaf4a163669933cab943ff","unresolved":true,"context_lines":[{"line_number":154,"context_line":""},{"line_number":155,"context_line":"    @_port_filter_lock"},{"line_number":156,"context_line":"    def _apply_port_filter(self, device_ids, update_filter\u003dFalse):"},{"line_number":157,"context_line":"        step \u003d cfg.CONF.agent_resources_processing_step"},{"line_number":158,"context_line":"        devices \u003d {}"},{"line_number":159,"context_line":"        security_groups \u003d {}"},{"line_number":160,"context_line":"        security_group_member_ips \u003d {}"}],"source_content_type":"text/x-python","patch_set":3,"id":"4738a1a0_6d9bcc7d","line":157,"range":{"start_line":157,"start_character":15,"end_line":157,"end_character":55},"in_reply_to":"0958f2a0_487afda0","updated":"2021-07-30 10:03:40.000000000","message":"Your logic is completely right.\nBut goal of the patch was to give user a chance to adjust it if user needs it or wants to play with it for some reason.\nBy default nothing is changed.","commit_id":"00c2199b426e9123aafb5c7adad6f4a0c47e91bd"},{"author":{"_account_id":33145,"name":"Kevin Li","email":"liwenhai3@huawei.com","username":"liwenhai"},"change_message_id":"3c11a8e5ae9f323f0c9b2ea94d433b5386b0561e","unresolved":true,"context_lines":[{"line_number":155,"context_line":""},{"line_number":156,"context_line":"    @_port_filter_lock"},{"line_number":157,"context_line":"    def _apply_port_filter(self, device_ids, update_filter\u003dFalse):"},{"line_number":158,"context_line":"        step \u003d common_constants.AGENT_RES_PROCESSING_STEP"},{"line_number":159,"context_line":"        devices \u003d {}"},{"line_number":160,"context_line":"        security_groups \u003d {}"},{"line_number":161,"context_line":"        security_group_member_ips \u003d {}"}],"source_content_type":"text/x-python","patch_set":9,"id":"1ed48263_024ddf11","side":"PARENT","line":158,"range":{"start_line":158,"start_character":15,"end_line":158,"end_character":57},"updated":"2021-08-12 07:48:32.000000000","message":"RPC_RES_PROCESSING_STEP is used in another place neutron/agent/common/ovs_lib.py 497.\n\nThe value is to devide flow strings for ovs. IMO, we can keep two options for two differrent actions. Not sure why this constants is used for two differrent scenario.","commit_id":"476bf747832694faee92a28fc86a04209da7eca9"},{"author":{"_account_id":32927,"name":"mitya-eremeev-2","display_name":"Mitya Eremeev","email":"mitossvyaz@mail.ru","username":"mitya-eremeev-2"},"change_message_id":"0671ddd29eb56729c1ba456db490bcf77f557491","unresolved":true,"context_lines":[{"line_number":155,"context_line":""},{"line_number":156,"context_line":"    @_port_filter_lock"},{"line_number":157,"context_line":"    def _apply_port_filter(self, device_ids, update_filter\u003dFalse):"},{"line_number":158,"context_line":"        step \u003d common_constants.AGENT_RES_PROCESSING_STEP"},{"line_number":159,"context_line":"        devices \u003d {}"},{"line_number":160,"context_line":"        security_groups \u003d {}"},{"line_number":161,"context_line":"        security_group_member_ips \u003d {}"}],"source_content_type":"text/x-python","patch_set":9,"id":"c1f0be98_42651b97","side":"PARENT","line":158,"range":{"start_line":158,"start_character":15,"end_line":158,"end_character":57},"in_reply_to":"1ed48263_024ddf11","updated":"2021-08-12 14:40:53.000000000","message":"done","commit_id":"476bf747832694faee92a28fc86a04209da7eca9"}],"neutron/api/rpc/handlers/securitygroups_rpc.py":[{"author":{"_account_id":5948,"name":"Oleg Bondarev","email":"obondarev@mirantis.com","username":"obondarev"},"change_message_id":"a09ee7c39ed529489ad7a641ab7969444485b8c7","unresolved":true,"context_lines":[{"line_number":383,"context_line":"        for index in range(0, len(remote_group_ids), batch_size):"},{"line_number":384,"context_line":"            security_group_batch \u003d remote_group_ids[index: index + batch_size]"},{"line_number":385,"context_line":"            filters \u003d {\"security_group_ids\": security_group_batch}"},{"line_number":386,"context_line":"            for p in self.rcache.get_resources(\u0027Port\u0027, filters):"},{"line_number":387,"context_line":"                allowed_ips \u003d [(str(addr.ip_address), str(addr.mac_address))"},{"line_number":388,"context_line":"                               for addr in p.allowed_address_pairs]"},{"line_number":389,"context_line":"                port_ips \u003d [(str(addr.ip_address), str(p.mac_address))"}],"source_content_type":"text/x-python","patch_set":16,"id":"b29cefcc_dd716805","line":386,"range":{"start_line":386,"start_character":21,"end_line":386,"end_character":63},"updated":"2021-08-31 11:33:26.000000000","message":"No RPC communication happens at this step - all ports are already in local cache - thanks to push notifications framework. So I don\u0027t see any reason for batching here + it\u0027s definitely not an RPC batching","commit_id":"4c9a0b19573d9c070c02fd40416accebe948ac5c"},{"author":{"_account_id":32927,"name":"mitya-eremeev-2","display_name":"Mitya Eremeev","email":"mitossvyaz@mail.ru","username":"mitya-eremeev-2"},"change_message_id":"95a8ee5b178ef6cd37410cf874955f42d5001a6a","unresolved":true,"context_lines":[{"line_number":383,"context_line":"        for index in range(0, len(remote_group_ids), batch_size):"},{"line_number":384,"context_line":"            security_group_batch \u003d remote_group_ids[index: index + batch_size]"},{"line_number":385,"context_line":"            filters \u003d {\"security_group_ids\": security_group_batch}"},{"line_number":386,"context_line":"            for p in self.rcache.get_resources(\u0027Port\u0027, filters):"},{"line_number":387,"context_line":"                allowed_ips \u003d [(str(addr.ip_address), str(addr.mac_address))"},{"line_number":388,"context_line":"                               for addr in p.allowed_address_pairs]"},{"line_number":389,"context_line":"                port_ips \u003d [(str(addr.ip_address), str(p.mac_address))"}],"source_content_type":"text/x-python","patch_set":16,"id":"ce9fa0db_c643b867","line":386,"range":{"start_line":386,"start_character":21,"end_line":386,"end_character":63},"in_reply_to":"b29cefcc_dd716805","updated":"2021-09-01 07:20:29.000000000","message":"reverted back","commit_id":"4c9a0b19573d9c070c02fd40416accebe948ac5c"}],"neutron/conf/agent/database/agents_db.py":[{"author":{"_account_id":16688,"name":"Rodolfo Alonso","email":"ralonsoh@redhat.com","username":"rodolfo-alonso-hernandez"},"change_message_id":"2ba9ffab3a6eb1c867593dda400b10b459e71bea","unresolved":true,"context_lines":[{"line_number":42,"context_line":"                       \"enable_new_agents\u003dFalse. In the case, user\u0027s \""},{"line_number":43,"context_line":"                       \"resources will not be scheduled automatically to the \""},{"line_number":44,"context_line":"                       \"agent until admin changes admin_state_up to True.\")),"},{"line_number":45,"context_line":"    cfg.IntOpt(\"agent_resources_processing_step\","},{"line_number":46,"context_line":"               default\u003d_constants.AGENT_RES_PROCESSING_STEP, min\u003d1,"},{"line_number":47,"context_line":"               help\u003d_(\"Number of resources for neutron agent side functions \""},{"line_number":48,"context_line":"                      \"to deal with large sets.\")),"}],"source_content_type":"text/x-python","patch_set":3,"id":"ff892274_a76a27dd","line":45,"updated":"2021-07-29 09:56:56.000000000","message":"I understand the problem you present in the bug and this could mitigate it without increasing unnecessarily the RPC timeouts.\n\nWhat I don\u0027t know if this will accepted because we are making configurable something related to the internal agent processing.\n\nIn any case, if there is no way to improve the server data collection, I\u0027m in favor of this idea.","commit_id":"00c2199b426e9123aafb5c7adad6f4a0c47e91bd"},{"author":{"_account_id":5948,"name":"Oleg Bondarev","email":"obondarev@mirantis.com","username":"obondarev"},"change_message_id":"7a9c05a5715a200a661766ebea159c19aeea4e8a","unresolved":true,"context_lines":[{"line_number":53,"context_line":"                      \"have  a lower timeout probability. And according to \""},{"line_number":54,"context_line":"                      \"the testing result, step size 100 can indeed cost \""},{"line_number":55,"context_line":"                      \"about 10% much more time than 500/1000. But such extra \""},{"line_number":56,"context_line":"                      \" time looks inevitably needed to be sacrificed for the \""},{"line_number":57,"context_line":"                      \"restart success rate. The best value can be determined \""},{"line_number":58,"context_line":"                      \"empirically in your environment.\""},{"line_number":59,"context_line":"                      \"[1] http://paste.openstack.org/show/745685/\")),"}],"source_content_type":"text/x-python","patch_set":5,"id":"0a56d897_db275145","line":56,"range":{"start_line":56,"start_character":23,"end_line":56,"end_character":24},"updated":"2021-08-09 09:56:51.000000000","message":"extra space not needed","commit_id":"639e90fb5dbf8a05e9adba480a1be4caf1e0b3c6"},{"author":{"_account_id":5948,"name":"Oleg Bondarev","email":"obondarev@mirantis.com","username":"obondarev"},"change_message_id":"7a9c05a5715a200a661766ebea159c19aeea4e8a","unresolved":true,"context_lines":[{"line_number":62,"context_line":"               help\u003d_(\"Number of resources for neutron to divide \""},{"line_number":63,"context_line":"                      \"the large RPC call data sets. It can be reduced \""},{"line_number":64,"context_line":"                      \"if RPC timeout occurred. The best value can be \""},{"line_number":65,"context_line":"                      \" determined empirically in your environment.\"))"},{"line_number":66,"context_line":"]"},{"line_number":67,"context_line":""},{"line_number":68,"context_line":""}],"source_content_type":"text/x-python","patch_set":5,"id":"66a4fd9a_b00b4c2b","line":65,"range":{"start_line":65,"start_character":23,"end_line":65,"end_character":24},"updated":"2021-08-09 09:56:51.000000000","message":"extra space not needed","commit_id":"639e90fb5dbf8a05e9adba480a1be4caf1e0b3c6"},{"author":{"_account_id":16688,"name":"Rodolfo Alonso","email":"ralonsoh@redhat.com","username":"rodolfo-alonso-hernandez"},"change_message_id":"99ca503c3f71f7d2847ed01c6bd56efd94080fbf","unresolved":true,"context_lines":[{"line_number":44,"context_line":"                       \"agent until admin changes admin_state_up to True.\")),"},{"line_number":45,"context_line":"    cfg.IntOpt(\"agent_resources_processing_step\","},{"line_number":46,"context_line":"               default\u003d_constants.AGENT_RES_PROCESSING_STEP, min\u003d1,"},{"line_number":47,"context_line":"               help\u003d_(\"Number of resources for neutron agent side functions \""},{"line_number":48,"context_line":"                      \"to deal with large sets. Setting this value does not \""},{"line_number":49,"context_line":"                      \"count on special conditions, it is just a human \""},{"line_number":50,"context_line":"                      \"countable or scalable number. [1] gives us the method \""},{"line_number":51,"context_line":"                      \"to test the scale issue. And we have tested the value \""},{"line_number":52,"context_line":"                      \"of 1000, 500, 200, 100. But for 100 ovs-agent will \""}],"source_content_type":"text/x-python","patch_set":7,"id":"ab2ff07f_a02b6051","line":49,"range":{"start_line":47,"start_character":22,"end_line":49,"end_character":72},"updated":"2021-08-10 09:49:24.000000000","message":"This description is not accurate or doesn\u0027t inform about where is used.\n\nThis value is the number of resources the agent will request from the server at once, in one single RPC call. Reducing this number will mitigate the RPC timeouts when requesting objects with large data sets. Currently this is used to limit the requested number of security groups per device.\n\n^^ that could be a description.","commit_id":"9e86e991b568901aedec154f003a3efec641b4e9"},{"author":{"_account_id":32927,"name":"mitya-eremeev-2","display_name":"Mitya Eremeev","email":"mitossvyaz@mail.ru","username":"mitya-eremeev-2"},"change_message_id":"356628a40e25d7cceb2a47c7ed3a2e0db266f5ca","unresolved":true,"context_lines":[{"line_number":44,"context_line":"                       \"agent until admin changes admin_state_up to True.\")),"},{"line_number":45,"context_line":"    cfg.IntOpt(\"agent_resources_processing_step\","},{"line_number":46,"context_line":"               default\u003d_constants.AGENT_RES_PROCESSING_STEP, min\u003d1,"},{"line_number":47,"context_line":"               help\u003d_(\"Number of resources for neutron agent side functions \""},{"line_number":48,"context_line":"                      \"to deal with large sets. Setting this value does not \""},{"line_number":49,"context_line":"                      \"count on special conditions, it is just a human \""},{"line_number":50,"context_line":"                      \"countable or scalable number. [1] gives us the method \""},{"line_number":51,"context_line":"                      \"to test the scale issue. And we have tested the value \""},{"line_number":52,"context_line":"                      \"of 1000, 500, 200, 100. But for 100 ovs-agent will \""}],"source_content_type":"text/x-python","patch_set":7,"id":"a8c87ebe_6c6b5808","line":49,"range":{"start_line":47,"start_character":22,"end_line":49,"end_character":72},"in_reply_to":"ab2ff07f_a02b6051","updated":"2021-08-11 14:02:34.000000000","message":"done","commit_id":"9e86e991b568901aedec154f003a3efec641b4e9"},{"author":{"_account_id":16688,"name":"Rodolfo Alonso","email":"ralonsoh@redhat.com","username":"rodolfo-alonso-hernandez"},"change_message_id":"99ca503c3f71f7d2847ed01c6bd56efd94080fbf","unresolved":true,"context_lines":[{"line_number":48,"context_line":"                      \"to deal with large sets. Setting this value does not \""},{"line_number":49,"context_line":"                      \"count on special conditions, it is just a human \""},{"line_number":50,"context_line":"                      \"countable or scalable number. [1] gives us the method \""},{"line_number":51,"context_line":"                      \"to test the scale issue. And we have tested the value \""},{"line_number":52,"context_line":"                      \"of 1000, 500, 200, 100. But for 100 ovs-agent will \""},{"line_number":53,"context_line":"                      \"have  a lower timeout probability. And according to \""},{"line_number":54,"context_line":"                      \"the testing result, step size 100 can indeed cost \""},{"line_number":55,"context_line":"                      \"about 10% much more time than 500/1000. But such extra \""},{"line_number":56,"context_line":"                      \"time looks inevitably needed to be sacrificed for the \""},{"line_number":57,"context_line":"                      \"restart success rate. The best value can be determined \""},{"line_number":58,"context_line":"                      \"empirically in your environment.\""},{"line_number":59,"context_line":"                      \"[1] http://paste.openstack.org/show/745685/\")),"},{"line_number":60,"context_line":"    cfg.IntOpt(\"rpc_resources_processing_step\","},{"line_number":61,"context_line":"               default\u003d_constants.RPC_RES_PROCESSING_STEP, min\u003d1,"}],"source_content_type":"text/x-python","patch_set":7,"id":"888207e3_fe21de66","line":58,"range":{"start_line":51,"start_character":20,"end_line":58,"end_character":56},"updated":"2021-08-10 09:49:24.000000000","message":"That justification should be in the release note, not in the option description.","commit_id":"9e86e991b568901aedec154f003a3efec641b4e9"},{"author":{"_account_id":32927,"name":"mitya-eremeev-2","display_name":"Mitya Eremeev","email":"mitossvyaz@mail.ru","username":"mitya-eremeev-2"},"change_message_id":"356628a40e25d7cceb2a47c7ed3a2e0db266f5ca","unresolved":true,"context_lines":[{"line_number":48,"context_line":"                      \"to deal with large sets. Setting this value does not \""},{"line_number":49,"context_line":"                      \"count on special conditions, it is just a human \""},{"line_number":50,"context_line":"                      \"countable or scalable number. [1] gives us the method \""},{"line_number":51,"context_line":"                      \"to test the scale issue. And we have tested the value \""},{"line_number":52,"context_line":"                      \"of 1000, 500, 200, 100. But for 100 ovs-agent will \""},{"line_number":53,"context_line":"                      \"have  a lower timeout probability. And according to \""},{"line_number":54,"context_line":"                      \"the testing result, step size 100 can indeed cost \""},{"line_number":55,"context_line":"                      \"about 10% much more time than 500/1000. But such extra \""},{"line_number":56,"context_line":"                      \"time looks inevitably needed to be sacrificed for the \""},{"line_number":57,"context_line":"                      \"restart success rate. The best value can be determined \""},{"line_number":58,"context_line":"                      \"empirically in your environment.\""},{"line_number":59,"context_line":"                      \"[1] http://paste.openstack.org/show/745685/\")),"},{"line_number":60,"context_line":"    cfg.IntOpt(\"rpc_resources_processing_step\","},{"line_number":61,"context_line":"               default\u003d_constants.RPC_RES_PROCESSING_STEP, min\u003d1,"}],"source_content_type":"text/x-python","patch_set":7,"id":"634c6d28_66b8fc20","line":58,"range":{"start_line":51,"start_character":20,"end_line":58,"end_character":56},"in_reply_to":"888207e3_fe21de66","updated":"2021-08-11 14:02:34.000000000","message":"done","commit_id":"9e86e991b568901aedec154f003a3efec641b4e9"},{"author":{"_account_id":5948,"name":"Oleg Bondarev","email":"obondarev@mirantis.com","username":"obondarev"},"change_message_id":"2f0ba24ac6354a7bce74b6d621e58d014a000463","unresolved":true,"context_lines":[{"line_number":42,"context_line":"                       \"enable_new_agents\u003dFalse. In the case, user\u0027s \""},{"line_number":43,"context_line":"                       \"resources will not be scheduled automatically to the \""},{"line_number":44,"context_line":"                       \"agent until admin changes admin_state_up to True.\")),"},{"line_number":45,"context_line":"    cfg.IntOpt(\"agent_resources_processing_step\","},{"line_number":46,"context_line":"               default\u003d_constants.AGENT_RES_PROCESSING_STEP, min\u003d1,"},{"line_number":47,"context_line":"               help\u003d_(\"This value is the number of resources the agent will \""},{"line_number":48,"context_line":"                      \"request from the server at once, in one single call. \""},{"line_number":49,"context_line":"                      \"Reducing this number will mitigate the timeouts when \""},{"line_number":50,"context_line":"                      \"requesting objects with large data sets. Currently \""},{"line_number":51,"context_line":"                      \"this is used to limit the requested number of security \""},{"line_number":52,"context_line":"                      \"groups per device.\")),"},{"line_number":53,"context_line":"    cfg.IntOpt(\"rpc_resources_processing_step\","},{"line_number":54,"context_line":"               default\u003d_constants.RPC_RES_PROCESSING_STEP, min\u003d1,"},{"line_number":55,"context_line":"               help\u003d_(\"Number of resources for neutron to divide \""},{"line_number":56,"context_line":"                      \"the large RPC call data sets. It can be reduced \""},{"line_number":57,"context_line":"                      \"if RPC timeout occurred. The best value can be \""},{"line_number":58,"context_line":"                      \"determined empirically in your environment.\"))"},{"line_number":59,"context_line":"]"},{"line_number":60,"context_line":""},{"line_number":61,"context_line":""}],"source_content_type":"text/x-python","patch_set":11,"id":"a19b8d97_f6960cee","line":58,"range":{"start_line":45,"start_character":4,"end_line":58,"end_character":69},"updated":"2021-08-16 07:52:29.000000000","message":"I know this has passed many iterations but still I\u0027m not sure I see the difference between these 2 options: both are about number of resources requested from server by RPC. Why not one single conf for this? Then OVS flow related const/conf could be added in LIU\u0027s patch.","commit_id":"22aa16e4e78be00f5ddde3041d799877ef0e3c7c"},{"author":{"_account_id":5948,"name":"Oleg Bondarev","email":"obondarev@mirantis.com","username":"obondarev"},"change_message_id":"6e1a68ba3370bb7ddfdad1d9d172021703f84112","unresolved":true,"context_lines":[{"line_number":42,"context_line":"                       \"enable_new_agents\u003dFalse. In the case, user\u0027s \""},{"line_number":43,"context_line":"                       \"resources will not be scheduled automatically to the \""},{"line_number":44,"context_line":"                       \"agent until admin changes admin_state_up to True.\")),"},{"line_number":45,"context_line":"    cfg.IntOpt(\"agent_resources_processing_step\","},{"line_number":46,"context_line":"               default\u003d_constants.AGENT_RES_PROCESSING_STEP, min\u003d1,"},{"line_number":47,"context_line":"               help\u003d_(\"This value is the number of resources the agent will \""},{"line_number":48,"context_line":"                      \"request from the server at once, in one single call. \""},{"line_number":49,"context_line":"                      \"Reducing this number will mitigate the timeouts when \""},{"line_number":50,"context_line":"                      \"requesting objects with large data sets. Currently \""},{"line_number":51,"context_line":"                      \"this is used to limit the requested number of security \""},{"line_number":52,"context_line":"                      \"groups per device.\")),"},{"line_number":53,"context_line":"    cfg.IntOpt(\"rpc_resources_processing_step\","},{"line_number":54,"context_line":"               default\u003d_constants.RPC_RES_PROCESSING_STEP, min\u003d1,"},{"line_number":55,"context_line":"               help\u003d_(\"Number of resources for neutron to divide \""},{"line_number":56,"context_line":"                      \"the large RPC call data sets. It can be reduced \""},{"line_number":57,"context_line":"                      \"if RPC timeout occurred. The best value can be \""},{"line_number":58,"context_line":"                      \"determined empirically in your environment.\"))"},{"line_number":59,"context_line":"]"},{"line_number":60,"context_line":""},{"line_number":61,"context_line":""}],"source_content_type":"text/x-python","patch_set":11,"id":"719aa25b_883d82e9","line":58,"range":{"start_line":45,"start_character":4,"end_line":58,"end_character":69},"in_reply_to":"085f6afd_dfce6c55","updated":"2021-08-16 08:34:22.000000000","message":"I\u0027d set it to 20 (default for RPC chunk size) - and IIRC your issue showed that 100 was too big for security-groups RPC. I believe 100 was added mostly for OVS flows processing.","commit_id":"22aa16e4e78be00f5ddde3041d799877ef0e3c7c"},{"author":{"_account_id":32927,"name":"mitya-eremeev-2","display_name":"Mitya Eremeev","email":"mitossvyaz@mail.ru","username":"mitya-eremeev-2"},"change_message_id":"e991fc5eb7f3be950ed737eb6d4286e3d18a4a1b","unresolved":true,"context_lines":[{"line_number":42,"context_line":"                       \"enable_new_agents\u003dFalse. In the case, user\u0027s \""},{"line_number":43,"context_line":"                       \"resources will not be scheduled automatically to the \""},{"line_number":44,"context_line":"                       \"agent until admin changes admin_state_up to True.\")),"},{"line_number":45,"context_line":"    cfg.IntOpt(\"agent_resources_processing_step\","},{"line_number":46,"context_line":"               default\u003d_constants.AGENT_RES_PROCESSING_STEP, min\u003d1,"},{"line_number":47,"context_line":"               help\u003d_(\"This value is the number of resources the agent will \""},{"line_number":48,"context_line":"                      \"request from the server at once, in one single call. \""},{"line_number":49,"context_line":"                      \"Reducing this number will mitigate the timeouts when \""},{"line_number":50,"context_line":"                      \"requesting objects with large data sets. Currently \""},{"line_number":51,"context_line":"                      \"this is used to limit the requested number of security \""},{"line_number":52,"context_line":"                      \"groups per device.\")),"},{"line_number":53,"context_line":"    cfg.IntOpt(\"rpc_resources_processing_step\","},{"line_number":54,"context_line":"               default\u003d_constants.RPC_RES_PROCESSING_STEP, min\u003d1,"},{"line_number":55,"context_line":"               help\u003d_(\"Number of resources for neutron to divide \""},{"line_number":56,"context_line":"                      \"the large RPC call data sets. It can be reduced \""},{"line_number":57,"context_line":"                      \"if RPC timeout occurred. The best value can be \""},{"line_number":58,"context_line":"                      \"determined empirically in your environment.\"))"},{"line_number":59,"context_line":"]"},{"line_number":60,"context_line":""},{"line_number":61,"context_line":""}],"source_content_type":"text/x-python","patch_set":11,"id":"085f6afd_dfce6c55","line":58,"range":{"start_line":45,"start_character":4,"end_line":58,"end_character":69},"in_reply_to":"a19b8d97_f6960cee","updated":"2021-08-16 08:21:30.000000000","message":"I can do one common config option, but these two options have different default values (100 and 20).\nWhat default value should I set for one common option ?","commit_id":"22aa16e4e78be00f5ddde3041d799877ef0e3c7c"}],"neutron/conf/agent/ovs_conf.py":[{"author":{"_account_id":9531,"name":"liuyulong","display_name":"LIU Yulong","email":"i@liuyulong.me","username":"LIU-Yulong"},"change_message_id":"4bfaf9a5dd71e08e43a2175aecf3bc2bbbef687c","unresolved":true,"context_lines":[{"line_number":44,"context_line":"                       \u0027unregistered multicast packets to all ports. \u0027"},{"line_number":45,"context_line":"                       \u0027The switch will send unregistered multicast packets \u0027"},{"line_number":46,"context_line":"                       \u0027only to ports connected to multicast routers.\u0027)),"},{"line_number":47,"context_line":"    cfg.IntOpt(\"ovs_flow_string_processing_step\","},{"line_number":48,"context_line":"               default\u003d100, min\u003d1,"},{"line_number":49,"context_line":"               help\u003d_(\"The value is to divide flow strings for ovs.\"))"},{"line_number":50,"context_line":"]"},{"line_number":51,"context_line":""},{"line_number":52,"context_line":""}],"source_content_type":"text/x-python","patch_set":10,"id":"358fce42_d906f893","line":49,"range":{"start_line":47,"start_character":0,"end_line":49,"end_character":70},"updated":"2021-08-13 15:40:07.000000000","message":"Same here, this is not needed.","commit_id":"9b530b0040624203862d1376a73e837663a4aeb5"}],"releasenotes/notes/introduced-config-options-for-agent-step-size-dd9595875de2b885.yaml":[{"author":{"_account_id":5948,"name":"Oleg Bondarev","email":"obondarev@mirantis.com","username":"obondarev"},"change_message_id":"754469e155fa1bde07ae1baa28f78d4817763da7","unresolved":true,"context_lines":[{"line_number":1,"context_line":"---"},{"line_number":2,"context_line":"fixes:"},{"line_number":3,"context_line":"  - |"},{"line_number":4,"context_line":"    Introduced 2 config options for agent step size customization:"},{"line_number":5,"context_line":"    1. agent_resources_processing_step - Number of resources for neutron agent"},{"line_number":6,"context_line":"       side functions to deal with large sets. Setting this value does not"},{"line_number":7,"context_line":"       count on special conditions, it is just a human countable or scalable"},{"line_number":8,"context_line":"       number. [1] gives us the method to test the scale issue. And we have"}],"source_content_type":"text/x-yaml","patch_set":8,"id":"0c2aa1e4_526f136f","line":5,"range":{"start_line":4,"start_character":4,"end_line":5,"end_character":78},"updated":"2021-08-11 11:31:41.000000000","message":"Indeed it rendered wrongly: https://351aa91bd4380912e93e-b4e19c98e06eb23bd3ecd1208eab0884.ssl.cf1.rackcdn.com/802596/8/check/build-openstack-releasenotes/35cd24b/docs/en_GB/unreleased.html\n\nCheck how it\u0027s done in releasenotes/notes/deprecate-designate-auth-opts-c2b1050a0360981c.yaml\n\nLooks like at least new line after #4 is needed","commit_id":"eb3588897ac4e4e83d8f811d662c369afa445e02"},{"author":{"_account_id":32927,"name":"mitya-eremeev-2","display_name":"Mitya Eremeev","email":"mitossvyaz@mail.ru","username":"mitya-eremeev-2"},"change_message_id":"356628a40e25d7cceb2a47c7ed3a2e0db266f5ca","unresolved":true,"context_lines":[{"line_number":1,"context_line":"---"},{"line_number":2,"context_line":"fixes:"},{"line_number":3,"context_line":"  - |"},{"line_number":4,"context_line":"    Introduced 2 config options for agent step size customization:"},{"line_number":5,"context_line":"    1. agent_resources_processing_step - Number of resources for neutron agent"},{"line_number":6,"context_line":"       side functions to deal with large sets. Setting this value does not"},{"line_number":7,"context_line":"       count on special conditions, it is just a human countable or scalable"},{"line_number":8,"context_line":"       number. [1] gives us the method to test the scale issue. And we have"}],"source_content_type":"text/x-yaml","patch_set":8,"id":"b309fe87_a2df6058","line":5,"range":{"start_line":4,"start_character":4,"end_line":5,"end_character":78},"in_reply_to":"0c2aa1e4_526f136f","updated":"2021-08-11 14:02:34.000000000","message":"It helped.\nThanks a lot !","commit_id":"eb3588897ac4e4e83d8f811d662c369afa445e02"}]}