| *** rossella_s has joined #openstack-neutron | 00:00 | |
| *** hoangcx has quit IRC | 00:01 | |
| *** salv-orlando has joined #openstack-neutron | 00:01 | |
| *** Swami has quit IRC | 00:02 | |
| *** ihrachys has quit IRC | 00:02 | |
| *** thorst_ has joined #openstack-neutron | 00:02 | |
| *** salv-orlando has quit IRC | 00:06 | |
| *** thorst_ has quit IRC | 00:10 | |
| *** elo has joined #openstack-neutron | 00:10 | |
| *** techcet has joined #openstack-neutron | 00:11 | |
| *** banix has joined #openstack-neutron | 00:15 | |
| openstackgerrit | Kevin Benton proposed openstack/neutron: Change external_network_bridge default to '' https://review.openstack.org/298443 | 00:16 |
|---|---|---|
| *** techcet has quit IRC | 00:17 | |
| *** techcet has joined #openstack-neutron | 00:20 | |
| *** kevo has quit IRC | 00:25 | |
| *** sputnik13 has quit IRC | 00:26 | |
| *** david-lyle_ has joined #openstack-neutron | 00:28 | |
| *** mfuruta has joined #openstack-neutron | 00:29 | |
| *** hieulq has quit IRC | 00:29 | |
| *** tbachman has quit IRC | 00:32 | |
| openstackgerrit | Rawlin Peters proposed openstack/neutron: Pass bridge_name in OVS trunk parent port's vif_details https://review.openstack.org/343816 | 00:32 |
| *** tbachman has joined #openstack-neutron | 00:37 | |
| *** emagana has joined #openstack-neutron | 00:37 | |
| openstackgerrit | Rawlin Peters proposed openstack/neutron: Pass bridge_name in OVS trunk parent port's vif_details https://review.openstack.org/343816 | 00:37 |
| *** techcet has quit IRC | 00:38 | |
| *** nlahouti has quit IRC | 00:38 | |
| *** emagana has quit IRC | 00:41 | |
| *** thorst_ has joined #openstack-neutron | 00:43 | |
| *** bigjools has quit IRC | 00:43 | |
| *** thorst_ has quit IRC | 00:44 | |
| *** bigjools has joined #openstack-neutron | 00:44 | |
| *** thorst_ has joined #openstack-neutron | 00:44 | |
| *** hieulq has joined #openstack-neutron | 00:45 | |
| *** namnh has joined #openstack-neutron | 00:46 | |
| *** hoangcx2 has quit IRC | 00:50 | |
| *** thorst_ has quit IRC | 00:52 | |
| *** mirrorbox has quit IRC | 00:59 | |
| *** gvrangan has quit IRC | 01:00 | |
| *** djan has quit IRC | 01:02 | |
| *** hoangcx has joined #openstack-neutron | 01:02 | |
| *** djan has joined #openstack-neutron | 01:02 | |
| *** zhhuabj has quit IRC | 01:03 | |
| *** stanzgy has joined #openstack-neutron | 01:04 | |
| *** tidwellr has left #openstack-neutron | 01:05 | |
| *** chlong has joined #openstack-neutron | 01:06 | |
| *** gvrangan has joined #openstack-neutron | 01:06 | |
| *** gouthamr has joined #openstack-neutron | 01:08 | |
| *** trananhkma has joined #openstack-neutron | 01:09 | |
| *** yamamoto_ has joined #openstack-neutron | 01:12 | |
| *** sindhu has joined #openstack-neutron | 01:13 | |
| *** david-lyle_ has quit IRC | 01:14 | |
| *** fnaval has quit IRC | 01:16 | |
| *** zhhuabj has joined #openstack-neutron | 01:16 | |
| *** Sukhdev has quit IRC | 01:18 | |
| openstackgerrit | Merged openstack/neutron: Pass timeout in milliseconds to timer_wait https://review.openstack.org/346973 | 01:20 |
| *** banix has quit IRC | 01:24 | |
| *** mirrorbox has joined #openstack-neutron | 01:25 | |
| *** mirrorbox has joined #openstack-neutron | 01:25 | |
| *** thorst_ has joined #openstack-neutron | 01:25 | |
| *** charliekang_ has joined #openstack-neutron | 01:25 | |
| *** charliekang has quit IRC | 01:26 | |
| openstackgerrit | Akihiro Motoki proposed openstack/python-neutronclient: Move find_resource family to API binding layer https://review.openstack.org/348096 | 01:26 |
| openstackgerrit | Akihiro Motoki proposed openstack/python-neutronclient: Add common utilities for OSC plugin implementation https://review.openstack.org/348097 | 01:26 |
| openstackgerrit | Akihiro Motoki proposed openstack/python-neutronclient: Implement OSC plugin version of VPN service https://review.openstack.org/348098 | 01:26 |
| *** mfuruta has quit IRC | 01:26 | |
| *** mfuruta has joined #openstack-neutron | 01:26 | |
| *** sdague has quit IRC | 01:29 | |
| *** shashank_hegde has quit IRC | 01:30 | |
| *** banix has joined #openstack-neutron | 01:30 | |
| *** charliekang_ has quit IRC | 01:31 | |
| *** thorst_ has quit IRC | 01:34 | |
| openstackgerrit | Abhishek Raut proposed openstack/python-neutronclient: Add trunk commands to openstackclient https://review.openstack.org/340624 | 01:35 |
| openstackgerrit | Aradhana Singh proposed openstack/neutron: Refactoring config options for cmd https://review.openstack.org/323046 | 01:36 |
| openstackgerrit | Kevin Benton proposed openstack/neutron: Add flavor/service provider support to routers https://review.openstack.org/268941 | 01:36 |
| *** abhiraut has quit IRC | 01:37 | |
| openstackgerrit | Lujin Luo proposed openstack/neutron: Add a unique key to port_id in routerports table https://review.openstack.org/285048 | 01:38 |
| *** djan has quit IRC | 01:41 | |
| *** tbachman_ has joined #openstack-neutron | 01:41 | |
| *** djan has joined #openstack-neutron | 01:43 | |
| *** tbachman has quit IRC | 01:43 | |
| *** tbachman_ is now known as tbachman | 01:43 | |
| *** techcet has joined #openstack-neutron | 01:44 | |
| *** sbalukoff has joined #openstack-neutron | 01:45 | |
| *** s3wong has quit IRC | 01:47 | |
| *** zhhuabj has quit IRC | 01:50 | |
| openstackgerrit | Armando Migliaccio proposed openstack/neutron: Add RPC layer for Trunk Plugin and initial Open vSwitch driver https://review.openstack.org/347662 | 01:53 |
| openstackgerrit | Jin Jing Lin proposed openstack/neutron-specs: IPsec traffic metering for VPNaaS https://review.openstack.org/346051 | 01:56 |
| *** annp has joined #openstack-neutron | 01:58 | |
| *** banix has quit IRC | 02:00 | |
| openstackgerrit | Irina proposed openstack/neutron-vpnaas: Adding tests for endpoint-group api. https://review.openstack.org/338075 | 02:02 |
| *** zhhuabj has joined #openstack-neutron | 02:03 | |
| *** thorst_ has joined #openstack-neutron | 02:04 | |
| *** baoli has quit IRC | 02:06 | |
| *** amotoki has quit IRC | 02:10 | |
| *** thorst_ has quit IRC | 02:10 | |
| *** banix has joined #openstack-neutron | 02:12 | |
| *** iyamahat_ has quit IRC | 02:12 | |
| *** yamahata has quit IRC | 02:13 | |
| *** ajmiller has joined #openstack-neutron | 02:13 | |
| *** sbalukoff has quit IRC | 02:16 | |
| *** sbalukoff has joined #openstack-neutron | 02:17 | |
| *** gvrangan has quit IRC | 02:18 | |
| *** itisha has quit IRC | 02:20 | |
| *** ajmiller has quit IRC | 02:22 | |
| *** tflynn has joined #openstack-neutron | 02:24 | |
| *** tflynn has quit IRC | 02:28 | |
| *** yamamoto_ has quit IRC | 02:31 | |
| *** amotoki has joined #openstack-neutron | 02:32 | |
| *** amotoki has quit IRC | 02:36 | |
| *** jinli has quit IRC | 02:37 | |
| *** xenogear has quit IRC | 02:38 | |
| *** zgleason has quit IRC | 02:38 | |
| *** thorst_ has joined #openstack-neutron | 02:40 | |
| *** amotoki has joined #openstack-neutron | 02:40 | |
| *** thorst_ has quit IRC | 02:41 | |
| openstackgerrit | Aradhana Singh proposed openstack/neutron: Refactoring config options for agent/common/ovs_lib https://review.openstack.org/346567 | 02:41 |
| *** jinli has joined #openstack-neutron | 02:42 | |
| *** yuanying has quit IRC | 02:47 | |
| *** gvrangan has joined #openstack-neutron | 02:48 | |
| *** xenogear has joined #openstack-neutron | 02:53 | |
| *** tbachman_ has joined #openstack-neutron | 02:54 | |
| *** agireud has quit IRC | 02:55 | |
| *** gouthamr_ has joined #openstack-neutron | 02:55 | |
| *** tbachman has quit IRC | 02:55 | |
| *** tbachman_ is now known as tbachman | 02:55 | |
| openstackgerrit | LIU Yulong proposed openstack/neutron: Clean up L3 agent side HA router stuffs https://review.openstack.org/265672 | 02:55 |
| *** zgleason has joined #openstack-neutron | 02:57 | |
| *** agireud has joined #openstack-neutron | 02:58 | |
| *** gouthamr has quit IRC | 02:59 | |
| *** amotoki has quit IRC | 03:01 | |
| openstackgerrit | QunyingRan proposed openstack/neutron: Remove used ip which not in subnet pool from network used ip statistics https://review.openstack.org/322387 | 03:06 |
| *** techcet has quit IRC | 03:06 | |
| openstackgerrit | Armando Migliaccio proposed openstack/neutron: Add RPC layer for Trunk Plugin and initial Open vSwitch driver https://review.openstack.org/347662 | 03:07 |
| *** amotoki has joined #openstack-neutron | 03:10 | |
| *** gouthamr_ has quit IRC | 03:10 | |
| *** vishwanathj has joined #openstack-neutron | 03:11 | |
| *** gvrangan has quit IRC | 03:12 | |
| *** fnaval has joined #openstack-neutron | 03:13 | |
| *** fnaval has quit IRC | 03:14 | |
| namnh | trananhkma: ping | 03:14 |
| *** gouthamr has joined #openstack-neutron | 03:14 | |
| *** fnaval has joined #openstack-neutron | 03:14 | |
| namnh | trananhkma: ping | 03:15 |
| *** tonytan4ever has joined #openstack-neutron | 03:17 | |
| *** tflynn has joined #openstack-neutron | 03:18 | |
| *** emagana has joined #openstack-neutron | 03:19 | |
| *** salv-orlando has joined #openstack-neutron | 03:20 | |
| *** sputnik13 has joined #openstack-neutron | 03:22 | |
| openstackgerrit | LIU Yulong proposed openstack/neutron: Ensure HA router can be synced after HA router race conditions https://review.openstack.org/265680 | 03:22 |
| *** tflynn has quit IRC | 03:22 | |
| *** gvrangan has joined #openstack-neutron | 03:22 | |
| *** emagana has quit IRC | 03:24 | |
| *** lujinluo has quit IRC | 03:25 | |
| *** shz has joined #openstack-neutron | 03:26 | |
| *** gvrangan has quit IRC | 03:29 | |
| openstackgerrit | Nam Nguyen Hoai proposed openstack/neutron: Preventing overlap CIDR on *one* network in case neutron active-active https://review.openstack.org/314054 | 03:29 |
| *** shihanzhang has quit IRC | 03:29 | |
| *** salv-orlando has quit IRC | 03:30 | |
| *** yatin has joined #openstack-neutron | 03:32 | |
| openstackgerrit | LIU Yulong proposed openstack/neutron: Ensure HA router can be synced after HA router race conditions https://review.openstack.org/265680 | 03:32 |
| *** sputnik13 has quit IRC | 03:33 | |
| *** mickeys has quit IRC | 03:33 | |
| *** mickeys has joined #openstack-neutron | 03:34 | |
| *** mickeys has quit IRC | 03:38 | |
| *** anilvenkata has joined #openstack-neutron | 03:40 | |
| *** anilvenkata has quit IRC | 03:44 | |
| *** bvandewa has quit IRC | 03:44 | |
| *** david-lyle_ has joined #openstack-neutron | 03:45 | |
| *** yatin has quit IRC | 03:46 | |
| *** yuanying has joined #openstack-neutron | 03:50 | |
| *** david-lyle_ is now known as david-lyle | 03:54 | |
| *** z_kassab has joined #openstack-neutron | 03:55 | |
| *** diga has joined #openstack-neutron | 03:56 | |
| *** liusheng has joined #openstack-neutron | 03:58 | |
| *** garyk1 has joined #openstack-neutron | 03:58 | |
| *** lujinluo has joined #openstack-neutron | 03:58 | |
| *** julim has quit IRC | 03:59 | |
| *** chandankumar has joined #openstack-neutron | 04:01 | |
| *** julim has joined #openstack-neutron | 04:02 | |
| *** julim has quit IRC | 04:02 | |
| *** abhiraut has joined #openstack-neutron | 04:03 | |
| *** links has joined #openstack-neutron | 04:03 | |
| *** david-lyle has quit IRC | 04:07 | |
| *** kevo has joined #openstack-neutron | 04:09 | |
| *** mkolesni has joined #openstack-neutron | 04:10 | |
| *** david-lyle has joined #openstack-neutron | 04:11 | |
| *** tflynn has joined #openstack-neutron | 04:12 | |
| *** pbandark has joined #openstack-neutron | 04:12 | |
| *** lizk has joined #openstack-neutron | 04:14 | |
| *** Sukhdev has joined #openstack-neutron | 04:15 | |
| *** tflynn has quit IRC | 04:16 | |
| *** wolverineav has quit IRC | 04:19 | |
| *** techcet has joined #openstack-neutron | 04:20 | |
| *** buttercup has joined #openstack-neutron | 04:25 | |
| *** fnaval has quit IRC | 04:25 | |
| *** wolverineav has joined #openstack-neutron | 04:26 | |
| openstackgerrit | Abhishek Raut proposed openstack/python-neutronclient: Add trunk commands to openstackclient https://review.openstack.org/340624 | 04:29 |
| *** tflynn has joined #openstack-neutron | 04:31 | |
| *** wolverineav has quit IRC | 04:31 | |
| *** gouthamr has quit IRC | 04:32 | |
| *** zhhuabj has quit IRC | 04:32 | |
| openstackgerrit | Darren Chan proposed openstack/neutron: Add link in README.rst https://review.openstack.org/348117 | 04:33 |
| *** abhiraut has quit IRC | 04:34 | |
| *** javeriak has joined #openstack-neutron | 04:36 | |
| *** javeriak has quit IRC | 04:38 | |
| *** shausy has joined #openstack-neutron | 04:39 | |
| *** tflynn has quit IRC | 04:41 | |
| *** pgadiya has joined #openstack-neutron | 04:43 | |
| *** jhershbe has joined #openstack-neutron | 04:45 | |
| *** zhhuabj has joined #openstack-neutron | 04:46 | |
| *** javeriak has joined #openstack-neutron | 04:48 | |
| *** liuyulong has joined #openstack-neutron | 04:49 | |
| *** shashank_hegde has joined #openstack-neutron | 04:51 | |
| *** banix has quit IRC | 04:55 | |
| *** muawiakhan has joined #openstack-neutron | 04:55 | |
| *** ratailor has joined #openstack-neutron | 04:57 | |
| *** javeriak has quit IRC | 04:57 | |
| *** Alex_Staf has joined #openstack-neutron | 04:57 | |
| *** Sukhdev has quit IRC | 04:58 | |
| *** z_kassab has quit IRC | 04:59 | |
| *** nlahouti has joined #openstack-neutron | 05:03 | |
| *** kevo has quit IRC | 05:06 | |
| *** hynekm has joined #openstack-neutron | 05:07 | |
| *** fnaval has joined #openstack-neutron | 05:09 | |
| *** zhhuabj has quit IRC | 05:09 | |
| *** fnaval has quit IRC | 05:10 | |
| *** fnaval has joined #openstack-neutron | 05:11 | |
| *** shashank_hegde has quit IRC | 05:11 | |
| *** abhiraut has joined #openstack-neutron | 05:12 | |
| *** lizk has quit IRC | 05:13 | |
| *** gvrangan has joined #openstack-neutron | 05:19 | |
| *** sridharg has joined #openstack-neutron | 05:20 | |
| *** ramishra_ has quit IRC | 05:22 | |
| *** kevo has joined #openstack-neutron | 05:23 | |
| *** ramishra has joined #openstack-neutron | 05:24 | |
| *** javeriak has joined #openstack-neutron | 05:25 | |
| *** javeriak has quit IRC | 05:25 | |
| *** javeriak has joined #openstack-neutron | 05:25 | |
| *** ekuris has joined #openstack-neutron | 05:27 | |
| *** oanson has joined #openstack-neutron | 05:27 | |
| *** abhiraut has quit IRC | 05:29 | |
| *** abhiraut has joined #openstack-neutron | 05:30 | |
| *** zhhuabj has joined #openstack-neutron | 05:31 | |
| *** abhiraut has quit IRC | 05:31 | |
| *** kevo has quit IRC | 05:31 | |
| *** bks has joined #openstack-neutron | 05:31 | |
| *** chandankumar has quit IRC | 05:32 | |
| *** chandankumar has joined #openstack-neutron | 05:32 | |
| *** abhiraut has joined #openstack-neutron | 05:33 | |
| *** abregman_ has joined #openstack-neutron | 05:33 | |
| openstackgerrit | Akihiro Motoki proposed openstack/neutron: bug tag: Add 'api-ref' for API reference https://review.openstack.org/348126 | 05:34 |
| *** techcet has quit IRC | 05:34 | |
| jschwarz | kevinbenton, ping | 05:35 |
| jschwarz | kevinbenton, re: https://review.openstack.org/#/c/323232/ | 05:36 |
| *** vikram has joined #openstack-neutron | 05:36 | |
| *** lujinluo has quit IRC | 05:36 | |
| *** lujinluo has joined #openstack-neutron | 05:36 | |
| *** javeriak_ has joined #openstack-neutron | 05:37 | |
| *** Alex_Staf has quit IRC | 05:37 | |
| *** iyamahat has joined #openstack-neutron | 05:37 | |
| *** anilvenkata has joined #openstack-neutron | 05:38 | |
| *** javeriak has quit IRC | 05:38 | |
| *** nyechiel has joined #openstack-neutron | 05:38 | |
| *** yfried has joined #openstack-neutron | 05:41 | |
| *** gongysh has joined #openstack-neutron | 05:41 | |
| *** abhiraut has quit IRC | 05:42 | |
| *** abhiraut has joined #openstack-neutron | 05:42 | |
| *** iyamahat has quit IRC | 05:43 | |
| *** kobis has joined #openstack-neutron | 05:43 | |
| *** kawa2014 has joined #openstack-neutron | 05:44 | |
| *** kobis has quit IRC | 05:46 | |
| garyk1 | amotoki: https://review.openstack.org/348121 - addressed your comments from the previous review | 05:46 |
| *** fnaval has quit IRC | 05:47 | |
| *** jprovazn has joined #openstack-neutron | 05:47 | |
| *** amotoki_ has joined #openstack-neutron | 05:48 | |
| *** amotoki has quit IRC | 05:51 | |
| *** mohankumar has joined #openstack-neutron | 05:51 | |
| *** kevo has joined #openstack-neutron | 05:52 | |
| *** mosulica has joined #openstack-neutron | 05:53 | |
| *** techcet has joined #openstack-neutron | 05:54 | |
| *** kevo has quit IRC | 05:54 | |
| *** mohankumar has quit IRC | 05:55 | |
| *** yamahata has joined #openstack-neutron | 05:55 | |
| *** gongysh has quit IRC | 05:58 | |
| openstackgerrit | Akihiro Motoki proposed openstack/neutron-lib: api-ref: Split LBaaS API reference into v1 and v2 https://review.openstack.org/348129 | 05:58 |
| openstackgerrit | Merged openstack/neutron: Add link in README.rst https://review.openstack.org/348117 | 05:58 |
| *** manikanta_tadi has joined #openstack-neutron | 06:01 | |
| openstackgerrit | Ankur proposed openstack/neutron: Neutron Feature Classification https://review.openstack.org/318192 | 06:02 |
| *** david-lyle has quit IRC | 06:02 | |
| *** abhiraut has quit IRC | 06:04 | |
| *** iwamoto has joined #openstack-neutron | 06:04 | |
| *** oshvartz has joined #openstack-neutron | 06:06 | |
| kevinbenton | jschwarz: pong | 06:07 |
| jschwarz | kevinbenton, good morning | 06:08 |
| jschwarz | kevinbenton, so I've discussed the ALLOCATING situation with amuller | 06:08 |
| kevinbenton | jschwarz: yeah | 06:08 |
| jschwarz | kevinbenton, the issue you described with CaS being "non-revertable" if a process has died before setting the router back to ACTIVE, also happens today | 06:09 |
| *** mohankumar has joined #openstack-neutron | 06:09 | |
| kevinbenton | jschwarz: how so? | 06:09 |
| kevinbenton | jschwarz: what blocks a scheduling process from picking it up? | 06:09 |
| jschwarz | kevinbenton, i.e. if a process sets the router to ALLOCATING, the only way (currently) to re-set it to ACTIVE is by setting its admin-state-up=False and then changing its ha state to False and back to True | 06:09 |
| kevinbenton | jschwarz: ah | 06:10 |
| jschwarz | kevinbenton, nothing blocks it - but nothing will change it back either | 06:10 |
| openstackgerrit | Swapnil Kulkarni (coolsvap) proposed openstack/neutron: [WIP] Testing latest u-c https://review.openstack.org/303346 | 06:10 |
| jschwarz | kevinbenton, I propose changing the code such that changing admin-state-up=False, also changes the status of the router to DOWN or something like that | 06:10 |
| kevinbenton | jschwarz: yeah, that sounds good | 06:10 |
| jschwarz | this will still require a manual operation from the user/admin, but it will allow us to go further with the CaS idea | 06:11 |
| jschwarz | :) | 06:11 |
| kevinbenton | jschwarz: i think we need to re-evaluate what protection ALLOCATING is offering | 06:11 |
| kevinbenton | jschwarz: is it just preventing stuff from getting to the agent? | 06:12 |
| jschwarz | kevinbenton, I agree, given the latest races we are seeing | 06:12 |
| jschwarz | kevinbenton, I think it is | 06:12 |
| jschwarz | the "concurrent schedule() and allocating" is still possible | 06:12 |
| jschwarz | this is what I was trying to fix in... | 06:12 |
| *** javeriak_ has quit IRC | 06:12 | |
| jschwarz | kevinbenton, in this: https://review.openstack.org/#/c/317949/ | 06:12 |
| *** kobis has joined #openstack-neutron | 06:13 | |
| jschwarz | (the entire schedule() and allocation is blocked on getting the ALLOCATING CaS, which is not the state today) | 06:13 |
| jschwarz | kevinbenton, specifically https://review.openstack.org/#/c/317949/13/neutron/scheduler/l3_agent_scheduler.py@56 takes care of schedule() only running for a single neutron-server thread | 06:14 |
| jschwarz | s/thread/worker/g | 06:14 |
| kevinbenton | jschwarz: why does an agent calling get_routers trigger scheduling? | 06:14 |
| kevinbenton | jschwarz: i can't remember why we have all of these paths | 06:14 |
| jschwarz | kevinbenton, auto_schedule_routers is called by get_routers | 06:15 |
| openstackgerrit | LIU Yulong proposed openstack/neutron-dynamic-routing: Remove duplicate test l3plugin https://review.openstack.org/347718 | 06:15 |
| kevinbenton | jschwarz: right, didn't amuller talk about getting rid of that? | 06:15 |
| kevinbenton | jschwarz: or am i thinking of something else | 06:15 |
| jschwarz | kevinbenton, https://github.com/openstack/neutron/blob/master/neutron/api/rpc/handlers/l3_rpc.py#L66 | 06:15 |
| jschwarz | kevinbenton, you're thinking of getting rid of it in the sync_routers call, I think | 06:15 |
| jschwarz | which we did a few weeks back :) | 06:16 |
| kevinbenton | jschwarz: ok, why don't we get rid of it here too? | 06:16 |
| kevinbenton | jschwarz: what purpose does it serve | 06:16 |
| jschwarz | kevinbenton, the auto_schedule_routers feature is the one that allows an l3 agent to ask for rescheduling | 06:16 |
| kevinbenton | jschwarz: you mean rescheduling on failures? | 06:17 |
| kevinbenton | jschwarz: *on agent failures | 06:17 |
| jschwarz | i.e., say a new l3 agent is added to a deployment - the auto_schedule_routers makes it so that Neutron will schedule ha routers to it since it can "carry the load" | 06:17 |
| jschwarz | kevinbenton, rescheduling on failures is another codepath i believe | 06:17 |
| *** pcaruana has joined #openstack-neutron | 06:17 | |
| kevinbenton | jschwarz: yeah, rescheduling is separate | 06:17 |
| kevinbenton | jschwarz: that's why i was asking | 06:17 |
| kevinbenton | jschwarz: so isn't this disruptive? | 06:17 |
| kevinbenton | jschwarz: bringing a new agent online moving routers around | 06:18 |
| jschwarz | kevinbenton, it is | 06:18 |
| jschwarz | kevinbenton, that's why https://review.openstack.org/#/c/317949/13/neutron/scheduler/l3_agent_scheduler.py@56 is proposed :) | 06:18 |
| jschwarz | kevinbenton, also, this only triggers for under-scheduled routers | 06:18 |
| jschwarz | kevinbenton, i.e. if a router is not HA it won't get scheduled to a new agent "just because" | 06:19 |
| kevinbenton | jschwarz: ok | 06:19 |
| *** gongysh has joined #openstack-neutron | 06:19 | |
| jschwarz | kevinbenton, HA routers will get re-scheduled, but then the new node will be in standby (and not master), which doesn't pose a disruption | 06:19 |
| kevinbenton | jschwarz: that makes more sense | 06:19 |
| jschwarz | :) | 06:19 |
| openstackgerrit | LIU Yulong proposed openstack/neutron: Ensure HA router can be synced after HA router race conditions https://review.openstack.org/265680 | 06:20 |
| jschwarz | kevinbenton, so I'm gonna get to work (an hour or so) and start working on these patches | 06:20 |
| jschwarz | propose a "revert ALLOCATING with admin-state-up" and re-opening the CaS, etc | 06:20 |
| jschwarz | kevinbenton, sounds good | 06:20 |
| jschwarz | ? | 06:21 |
| kevinbenton | jschwarz: i would feel a bit better about that since there is at least an admin path to undo it | 06:21 |
| jschwarz | excellent | 06:21 |
| jschwarz | pleasure doing business with you XD | 06:21 |
| kevinbenton | jschwarz: no prob :) | 06:21 |
| kevinbenton | jschwarz: https://review.openstack.org/#/c/265680/19/neutron/db/l3_hamode_db.py | 06:23 |
| kevinbenton | jschwarz: for example it bothers me that we still have to pepper these fixes | 06:23 |
| kevinbenton | jschwarz: but i think the issue here is that the router is retrieved before it changes status | 06:23 |
| kevinbenton | jschwarz: and then get_port is called afterwards for the ha interface | 06:24 |
| kevinbenton | jschwarz: when it could already be gone | 06:24 |
| kevinbenton | ajo: you online? | 06:26 |
| *** yamamoto_ has joined #openstack-neutron | 06:27 | |
| *** muawiakhan has quit IRC | 06:31 | |
| *** gongysh has quit IRC | 06:32 | |
| *** itzikb has joined #openstack-neutron | 06:35 | |
| *** vthapar has joined #openstack-neutron | 06:36 | |
| *** mickeys has joined #openstack-neutron | 06:36 | |
| *** tonytan4ever has quit IRC | 06:38 | |
| *** sridharg has quit IRC | 06:40 | |
| *** mickeys has quit IRC | 06:40 | |
| *** abregman_ is now known as abregman | 06:41 | |
| *** tesseract- has joined #openstack-neutron | 06:42 | |
| *** techcet has quit IRC | 06:42 | |
| *** moshele has joined #openstack-neutron | 06:42 | |
| openstackgerrit | Nam Nguyen Hoai proposed openstack/neutron: Preventing overlap CIDR on *one* network in case neutron active-active https://review.openstack.org/314054 | 06:43 |
| openstackgerrit | venkata anil proposed openstack/neutron: multiple port bindings for HA ports https://review.openstack.org/323314 | 06:46 |
| ajo | hey kevinbenton | 06:47 |
| ajo | sort of :) | 06:47 |
| ajo | kevinbenton, any update on the ofports thing? | 06:48 |
| kevinbenton | ajo: i have a question about some OVO stuff | 06:48 |
| ajo | kevinbenton, of course | 06:48 |
| kevinbenton | ajo: nah, just waiting for terry's fix to go in | 06:48 |
| ajo | ah, ok :) | 06:48 |
| kevinbenton | ajo: then will watch for failures | 06:48 |
| *** salv-orlando has joined #openstack-neutron | 06:48 | |
| kevinbenton | ajo: it seems like we have nothing that filters fields when we create using the OVO pattern | 06:49 |
| *** vishwanathj has quit IRC | 06:49 | |
| kevinbenton | ajo: or at least i can't find it | 06:49 |
| ajo | to create or to fetch? | 06:49 |
| ajo | what do you mean by filter fields on create? | 06:49 |
| *** rubasov has joined #openstack-neutron | 06:49 | |
| kevinbenton | ajo: https://github.com/openstack/neutron/blob/master/neutron/services/qos/qos_plugin.py#L44-L58 | 06:50 |
| ajo | aha, do you mean, to check fields ? | 06:50 |
| kevinbenton | ajo: normally the response to a create request goes through a function called _make_port_dict or something like that | 06:50 |
| openstackgerrit | venkata anil proposed openstack/neutron: multiple port bindings for HA ports https://review.openstack.org/323314 | 06:50 |
| kevinbenton | ajo: which filters out any fields not in the API/extensions | 06:50 |
| kevinbenton | ajo: https://github.com/openstack/neutron/blob/master/neutron/db/db_base_plugin_common.py#L179-L197 | 06:51 |
| ajo | hmm, the existing fields will be updated with the values in the dict (and checked for validity) | 06:51 |
| ajo | but the non existing ones... may be we have a bug there to handle | 06:51 |
| kevinbenton | ajo: no, i mean there can be fields on the OVO | 06:51 |
| kevinbenton | ajo: that shouldn't be visible in the API | 06:52 |
| kevinbenton | ajo: because there is no extension loaded that supports them | 06:52 |
| ajo | ah, ok so @db_base_plugin_common.convert_result_to_dict should probably be filtering stuff out | 06:52 |
| ajo | based on which extensions are there | 06:52 |
| ajo | kevinbenton, can you give me an specific example? | 06:52 |
| kevinbenton | ajo: are QOS policies standard attrs yet? | 06:53 |
| ajo | I'm still a bit fuzzy on how extensions & OVOs play together, but I guess Ihar, arthur, rosella have a better picture of that | 06:53 |
| ajo | kevinbenton, all on policies is standard, yes, I mean, you have the service or not | 06:53 |
| ajo | if you don't have the service, you don't get the qos_policy_id on ports & networks | 06:54 |
| ajo | but that's handled by an ml2 extension, or by your plugin itself | 06:54 |
| ajo | via some helpers we provided | 06:54 |
| kevinbenton | ajo: what i mean is does the qos object inherit from the hasstandardattributes thingy? | 06:54 |
| ajo | kevinbenton, I may need to check, I probably havent tracked all the OVO patches lately | 06:55 |
| ajo | there's more I'm able to read | 06:55 |
| ajo | 1 sec | 06:55 |
| ajo | let me trace that | 06:55 |
| kevinbenton | oh, it doesn't look like it yet | 06:55 |
| *** jlanoux has joined #openstack-neutron | 06:56 | |
| kevinbenton | ajo: so one way to test would be to add another field to the qos policy object | 06:57 |
| *** javeriak has joined #openstack-neutron | 06:57 | |
| kevinbenton | ajo: you will see that in the create response | 06:57 |
| kevinbenton | ajo: even though there is no API definition for it | 06:57 |
| openstackgerrit | steve proposed openstack/neutron-dynamic-routing: Add bgpvpn and speaker binding https://review.openstack.org/341273 | 06:58 |
| *** jlanoux has quit IRC | 06:58 | |
| *** jlanoux has joined #openstack-neutron | 06:59 | |
| ajo | kevinbenton, ouch, got disconnected from VPN, I'm back, | 07:00 |
| ajo | so I see port model has the standard attributes thing | 07:00 |
| ajo | ah, ok, and in that case OVO base class adds the STANDARD_ATTRIBUTES fields | 07:00 |
| ajo | automatically | 07:00 |
| ajo | revision_number, description, created_at, updated_at | 07:00 |
| kevinbenton | right | 07:00 |
| *** nherciu has joined #openstack-neutron | 07:00 | |
| ajo | kevinbenton, is it acting up or something? | 07:00 |
| kevinbenton | the problem is that it will display those via the API | 07:01 |
| kevinbenton | even if there is no extension with an API definition for them | 07:01 |
| kevinbenton | during create | 07:01 |
| ajo | ahhaa | 07:02 |
| ajo | yikes, yes, you told me, see?, I'm here, but I'm not yet X) | 07:02 |
| kevinbenton | :) | 07:03 |
| ajo | caffeine blood level not high enough | 07:03 |
| *** irenab has joined #openstack-neutron | 07:03 | |
| openstackgerrit | Armando Migliaccio proposed openstack/neutron: Add RPC layer for Trunk Plugin and initial Open vSwitch driver https://review.openstack.org/347662 | 07:03 |
| ajo | hi armax morning/night | 07:03 |
| armax | ajo: hi | 07:03 |
| *** ushkalim has quit IRC | 07:03 | |
| ajo | kevinbenton, so we need to have some way of filtering those | 07:03 |
| armax | ajo: I actually have a question for you | 07:04 |
| kevinbenton | ajo: yeah | 07:04 |
| ajo | armax, shoot :) | 07:04 |
| ajo | or shot? | 07:04 |
| ajo | mhm | 07:04 |
| armax | ajo: trying to use the rpc callbacks registry | 07:04 |
| armax | ajo: what if I want to push a list of objects? | 07:05 |
| armax | ajo: can’t seem to see a way to do that | 07:05 |
| *** Fdaisuke_ has joined #openstack-neutron | 07:05 | |
| *** gvrangan has quit IRC | 07:05 | |
| armax | ajo: unless I push a ‘container’ object | 07:05 |
| *** namnh has quit IRC | 07:05 | |
| ajo | armax, hmmm, not possible at this moment, but it would be a good enhancement, I can work that out if you need it | 07:05 |
| armax | ajo: let me show you where I need it | 07:05 |
| armax | ajo: if you could get it done while I go to sleep that would be awesome :) | 07:05 |
| *** annp has quit IRC | 07:05 | |
| *** hoangcx has quit IRC | 07:05 | |
| ajo | armax, correct, if there's an object with sub-objects, it works (we use that currently in the qospolicies) | 07:05 |
| ajo | armax, I don't have tight day, so it's likely I can handle it | 07:06 |
| armax | ajo: see here https://review.openstack.org/#/c/347662/5/neutron/services/trunk/rpc/agent.py | 07:06 |
| armax | and here: | 07:06 |
| * ajo looks | 07:06 | |
| armax | https://review.openstack.org/#/c/347662/5/neutron/services/trunk/rpc/server.py | 07:06 |
| armax | the reason why it’s not desirable for us to send the entire container object is because we’re interested in the ‘diff' | 07:07 |
| ajo | armax, I also wanted to add multi pull (to pull a list of objects) | 07:07 |
| *** Fdaisuke has quit IRC | 07:07 | |
| ajo | aha | 07:07 |
| armax | ajo: whereas the container has the entire lot | 07:07 |
| armax | ajo: if you know what I mean | 07:07 |
| armax | I guess we could get away by pushing every single instance, but it’s chattier than I would like | 07:08 |
| *** hoangcx has joined #openstack-neutron | 07:08 | |
| ajo | armax, I see, as long as the agent information is consistent in the end, which I'm sure you're handling.. | 07:08 |
| ajo | hmm | 07:08 |
| armax | ajo: btw, if you could review the way I hooked up with the rpc callabacks | 07:08 |
| ajo | We may need to handle version of the agents, to make sure they're able to swallow lists of objects or not | 07:08 |
| armax | ajo: well, baby steps, baby steps :) | 07:09 |
| armax | ajo: true | 07:09 |
| ajo | and selectively send lists, or singles | 07:09 |
| ajo | but we have a mechanism to know agent's object versions, so | 07:09 |
| ajo | we could figure it out | 07:09 |
| armax | ajo: so do you think we might be better off building an ‘in-memory diff object’? | 07:09 |
| armax | as crazy as it sounds? | 07:09 |
| armax | ultimately the use case we have is this: | 07:10 |
| *** devvesa has joined #openstack-neutron | 07:10 | |
| ajo | hmm, the diff object could have the advantage of having some sort of hash of the old set, and if that doesn't match in the agent, then the agent can pull all back from the server | 07:10 |
| armax | a user ‘add/removes a child resource to/from a parent resource’, and then the server would want to notify the agent of the children being added/removed | 07:11 |
| ajo | but not sure if that's an issue here | 07:11 |
| ajo | armax, hmm | 07:11 |
| ajo | armax, this is a TrunkPort object and Port objects as sub-objects ? | 07:11 |
| * ajo looks at the code | 07:11 | |
| armax | ajo: ya, Trunk vs SubPorts | 07:11 |
| ajo | ah, ok subports | 07:12 |
| ajo | it's ok to send a delete of a subport , a port would be a different thing | 07:12 |
| armax | https://github.com/openstack/neutron/blob/master/neutron/objects/trunk.py | 07:12 |
| armax | so the agent is interested in the subports being affected | 07:12 |
| armax | rather than the whole trunk | 07:12 |
| ajo | hm | 07:12 |
| armax | that’s why I thought a bulky push would be great | 07:13 |
| *** permalac has joined #openstack-neutron | 07:13 | |
| ajo | ok, so if you send the trunk, you get the whole thing (as per list of objects field) | 07:13 |
| ajo | and | 07:13 |
| ajo | then you could send deletes or additions over SubPorts | 07:13 |
| ajo | I get the point | 07:14 |
| ajo | eventually you could have a TrunkPort with 1000 SubPorts | 07:14 |
| ajo | and sending that chunk of RPC could be suboptimal | 07:14 |
| ajo | Well, in fact | 07:14 |
| ajo | you would be sending | 07:14 |
| ajo | N! increasing size messages | 07:15 |
| armax | I suppose a trunk object can be potentially big | 07:15 |
| armax | however if the user operation only affects a fraction of it | 07:15 |
| ajo | T{S1}, T{S1, S2} , T{S1,S2,S3} ...... T{S1,S2,S3 .... SN) | 07:15 |
| armax | ideally I’d send just that rather than the whole lot | 07:15 |
| ajo | yes, that makes sense, or we're going to torture the agent | 07:16 |
| ajo | and the server, | 07:16 |
| armax | by sending the subports one by one, for instance | 07:16 |
| ajo | yeah | 07:16 |
| armax | ajo: so what do you reckon? | 07:16 |
| *** hieulq has quit IRC | 07:16 | |
| armax | ajo: I am fried for the day | 07:16 |
| ajo | you could send subports one by one, as separate objects, | 07:16 |
| ajo | armax, ack, | 07:17 |
| ajo | I'll think of the problem | 07:17 |
| ajo | just one question | 07:17 |
| armax | ajo: true, but that’s chattier than it could be | 07:17 |
| armax | ajo: shoot | 07:17 |
| ajo | on the API, are the subports added one by one? or does it let you add in bulk? | 07:17 |
| armax | ajo: I thought of sending subports one by one | 07:17 |
| armax | the rest API adds them in bulk | 07:17 |
| ajo | ahh, aha | 07:17 |
| ajo | ok, I see the point then | 07:17 |
| *** agireud has quit IRC | 07:17 | |
| ajo | ok, I'll add a multi push | 07:18 |
| ajo | and think of the logic | 07:18 |
| armax | https://github.com/openstack/neutron/blob/master/neutron/services/trunk/plugin.py#L175 | 07:18 |
| ajo | I'll try to get that during your night | 07:18 |
| *** sridharg has joined #openstack-neutron | 07:18 | |
| armax | ajo: if you look at https://github.com/openstack/neutron/blob/master/neutron/services/trunk/plugin.py#L208 | 07:18 |
| armax | the payload contains the added_subports | 07:19 |
| ajo | ahaa | 07:19 |
| armax | ajo: then in here https://review.openstack.org/#/c/347662/5/neutron/services/trunk/rpc/backend.py@80 | 07:19 |
| *** techcet has joined #openstack-neutron | 07:19 | |
| armax | we would like to send the list to the agent | 07:20 |
| armax | and here: | 07:20 |
| armax | https://review.openstack.org/#/c/347662/5/neutron/services/trunk/rpc/server.py@78 | 07:20 |
| ajo | ah, ok, you translate from one layer to another | 07:20 |
| armax | I raised the TODO because I realized that the bulk was not avaiable | 07:20 |
| ajo | that's similar to what kevinbenton was doing with the standard attrs | 07:20 |
| ajo | we should eventually unify them | 07:20 |
| ajo | armax, makes sense | 07:21 |
| *** ihrachys has joined #openstack-neutron | 07:21 | |
| armax | ajo: if you provide some feedback that would be great | 07:21 |
| ajo | ack, I will review this, and go to the implementation of list push | 07:22 |
| *** nmagnezi has joined #openstack-neutron | 07:22 | |
| armax | ajo: because we either fall back to sending objects one by one, which it’s not ideal, or abandon the idea of sending actual objects, which I don’t like either | 07:22 |
| ajo | armax, naah, if we can't do list now for some weird technical reason, let's do it in a 2nd baby step later, but it seems possible to me so far | 07:22 |
| armax | ajo: unless there’s some other clever way of dealing with this which I haven’t thought of | 07:23 |
| korzen_ | kevinbenton, so the extension and OVO is kind of a new story but from your experience is that you have to define all fields in object upfront | 07:23 |
| ajo | you get your RPC more simplified this way, and you don't need to care about the version history of RPC | 07:23 |
| armax | ajo: ack | 07:23 |
| ajo | armax, the in-memory diff could be another option | 07:23 |
| armax | ajo: right, dunno how that works pratically | 07:23 |
| ajo | but, that adds more work, while the list-push is probably beneficial to other areas | 07:23 |
| armax | ajo: right | 07:24 |
| *** techcet has quit IRC | 07:24 | |
| ajo | armax, those would just be objects with no DB mapping | 07:24 |
| armax | I figured the bulky push could be useful and less cumbersome for other use cases | 07:24 |
| korzen_ | kevinbenton, then you can define extent_port/subnet/network_dict and fill in the not standard fields, the standard fields like defined in SQL schema will be loaded by object DB API | 07:24 |
| ajo | just field validations, and eventually, as they evolve, make_obj_compatible methods to downgrade | 07:24 |
| ajo | armax, totally agreed | 07:24 |
| armax | ajo: do we have an example in the tree? | 07:25 |
| *** gongysh has joined #openstack-neutron | 07:25 | |
| korzen_ | kevinbenton, for filtering we should use the make_port/subnet/network_dict as you have mentioned | 07:25 |
| kevinbenton | korzen_: fields in the schema can belong to an extension | 07:25 |
| ajo | armax, hmm, I think we don't but let me see | 07:25 |
| *** Alex_Stef has joined #openstack-neutron | 07:25 | |
| ajo | ihrachys, , korzen_ do we have any neutron-object in tree with no DB mapping ? | 07:25 |
| armax | ajo: ok, so if you can spare a few minutes to make a suggestion on https://review.openstack.org/#/c/347662/ that would be great | 07:25 |
| ihrachys | ajo: rule type? | 07:25 |
| ajo | oh right | 07:25 |
| ajo | let me fetch the link to rule type | 07:26 |
| kevinbenton | korzen_: for example, create_at/updated_at are only processed if the timestamp service plugin is loaded | 07:26 |
| ajo | armax, : example of non-db-mapped: https://github.com/openstack/neutron/blob/master/neutron/objects/qos/rule_type.py | 07:26 |
| kevinbenton | korzen_: they shouldn't be visible in the API otherwise | 07:26 |
| *** agireud has joined #openstack-neutron | 07:26 | |
| ajo | note the NeutronObject instead of NeutronDbObject top class | 07:26 |
| ihrachys | armax: what's the context of the discussion? should I look at some specific patch? | 07:27 |
| ihrachys | is it trunks? | 07:27 |
| ajo | ihrachys, yeah, it's related to trunks | 07:27 |
| armax | ajo: though technically the object is a container of objects | 07:27 |
| armax | ihrachys: I was playing with the rpc callbacks | 07:27 |
| ajo | armax, was considering to send subports creation when those are added, instead of sending the whole trunk port | 07:27 |
| korzen_ | kevinbenton, yes, for get_subnet I'm using make_subnet_dict, and then the extend_subnet_dict should fill in the create_at/updated_at in result dict, taking values from OVO | 07:27 |
| ajo | which would be very verbose to the RPC, because | 07:28 |
| armax | and I wondered whether tehre was a way to push a list of objects of the same type rather than a single one | 07:28 |
| *** nlahouti has quit IRC | 07:28 | |
| ajo | it can get big | 07:28 |
| ajo | but push | 07:28 |
| ajo | has no ability to send lists (yet) | 07:28 |
| korzen_ | kevinbenton, but for RPC callback I see that it is not implemented, filtering the fields | 07:28 |
| ihrachys | we can add it if needed | 07:28 |
| ajo | so we were talking of adding list capability to .push | 07:28 |
| kevinbenton | korzen_: RPC callback i'm not concerned about | 07:28 |
| armax | ihrachys: well, we’re figuring out what’s the best course of action :) | 07:28 |
| ajo | we just need to keep track of which agents can handle lists, and that's it | 07:28 |
| kevinbenton | korzen_: just API responses | 07:29 |
| *** tonytan4ever has joined #openstack-neutron | 07:29 | |
| ihrachys | though not sure if it buys much and you should spend cycles on that optimization right now | 07:29 |
| korzen_ | kevinbenton, so in my opinion it should be fine for API | 07:29 |
| kevinbenton | korzen_: what is fine? | 07:29 |
| *** pece has joined #openstack-neutron | 07:29 | |
| korzen_ | kevinbenton, subnet API should be fine | 07:29 |
| armax | ihrachys: sending objects one by one especially if the list is big is not great | 07:30 |
| ajo | armax, ihrachys may be we should measure whow big a TrunkPort get's when serialized with SubPorts attached | 07:30 |
| armax | ihrachys: surely we can optimize later | 07:30 |
| ihrachys | armax: the objects sent are subports or trunks? | 07:30 |
| armax | ihrachys: subports | 07:30 |
| *** fzdarsky has joined #openstack-neutron | 07:30 | |
| kevinbenton | korzen_: but the pattern used by qos for example means it cannot be converted to a standard attr | 07:30 |
| ajo | armax, the other option could be sending the whole TrunkPort if we see it's not getting too big over the wire | 07:31 |
| armax | ihrachys: in the end I can see having bulky push is something we’d want, no? | 07:31 |
| ihrachys | armax: I don't think it will be too huge to swallow. and as I said, that could be postponed to next cycles when we feel the need to optimize. | 07:31 |
| ihrachys | armax: in theory, yes, it could have its use cases | 07:31 |
| kevinbenton | korzen_: and the trunk patch is now leaking stuff as well because it followed the qos pattern and trunk has standard attributes | 07:31 |
| armax | ihrachys: it’s not a matter of how big it is on the wire | 07:31 |
| ajo | armax, what's the expected scalability of Trunk ports in terms of SubPorts ? | 07:31 |
| armax | ajo: right now it could go up to 4K due to the vlan segmentation type supported | 07:31 |
| armax | ihrachys: if we send the whole trunk, the agent has to figure out what subports changed | 07:32 |
| *** akijak has left #openstack-neutron | 07:32 | |
| armax | ihrachys: doable, but not ideal | 07:32 |
| ajo | well, so we could get up to 1MB message for a 4K trunk port | 07:32 |
| kevinbenton | korzen_: https://github.com/openstack/neutron/blob/1750d757613d08cb16fa16a5c76d52408590f6c2/neutron/services/qos/qos_plugin.py#L44-L58 | 07:32 |
| ajo | (by eye) | 07:32 |
| ajo | probably a bit less | 07:33 |
| armax | ihrachys: because the agent would have to figure out the diff from local trunk snapshot | 07:33 |
| armax | ihrachys: it’s worth considering, at this point I am vetting options | 07:33 |
| kevinbenton | armax, ihrachys: frankly the agent isn't that big of a concern, iterating a 4k list is fast | 07:33 |
| kevinbenton | i would be more concerned about rabbit | 07:33 |
| ajo | yeah, rabbit would probably suffer | 07:34 |
| ajo | and worst | 07:34 |
| armax | kevinbenton: right my concern is the overall latency | 07:34 |
| *** hoangcx has quit IRC | 07:34 | |
| ajo | it wouldn't be a 4K trunk port | 07:34 |
| armax | between the time the user added/removed suports to a trunk | 07:34 |
| ajo | it would be several of those as the subports are added in bulk | 07:34 |
| armax | and the whole thing is expected to finish | 07:34 |
| armax | from desired to actual state | 07:34 |
| *** wolverineav has joined #openstack-neutron | 07:34 | |
| ajo | so may be 100, 200, 300 ... so up to 4000 | 07:34 |
| *** lezbar has joined #openstack-neutron | 07:34 | |
| ajo | ihrachys, armax let me give it a shoot to the list push, | 07:34 |
| ajo | and if it seems not possible, or technically complicated, we explore other options | 07:35 |
| jschwarz | kevinbenton, back now :) looking at what you wrote | 07:35 |
| armax | ajo: ok, no pressure, give it some thought | 07:35 |
| kevinbenton | armax: why can't we just push subports? | 07:35 |
| armax | ajo: right | 07:35 |
| armax | kevinbenton: because rpc push does not support bulks | 07:35 |
| armax | only individual objects | 07:35 |
| kevinbenton | armax: i think pushing individual subports is better than a giant trunk | 07:36 |
| ajo | If I can scope an initial patch during today, I can help you write an in-memory diff like SubPortBatch | 07:36 |
| ajo | or something like that | 07:36 |
| kevinbenton | armax: if it comes down to that | 07:36 |
| armax | kevinbenton: a user add/removes a list of subports | 07:36 |
| *** abregman_ has joined #openstack-neutron | 07:36 | |
| armax | kevinbenton: I would like the server to notify the agent of the list of subports being added/removed | 07:36 |
| armax | in one go | 07:36 |
| *** muawiakhan has joined #openstack-neutron | 07:36 | |
| ihrachys | kevinbenton: maybe we should not even expose subports as a field on the trunk then. | 07:36 |
| armax | rather having to enumerate over the list | 07:36 |
| *** muawiakhan has quit IRC | 07:36 | |
| * ajo feels like it's a good excuse to get his hands dirty with the rpc callbacks and do some other enhancements :P | 07:36 | |
| ihrachys | kevinbenton: just provide some Trunk methods to operate on them, like we do to attach qos policies to ports. we don't expose bindings as a field | 07:37 |
| kevinbenton | armax: right, i'm just suggesting that i would prefer individual subports be pushed over the entire trunk | 07:37 |
| kevinbenton | armax: if the list thing doesn't work out | 07:37 |
| ihrachys | then your trunk object .to_dict() will be slim for rpc payload | 07:37 |
| kevinbenton | ihrachys: yeah | 07:37 |
| armax | kevinbenton: ok, I am not contemplating sending the entire trunk if that’s what you’re worried about :) | 07:37 |
| *** abregman has quit IRC | 07:37 | |
| ajo | ihrachys, tjat's a good point | 07:37 |
| ihrachys | ok guys sorry I need to run for an hour or so. please update me via email or whatnot if you have smth for me | 07:38 |
| ajo | armax, yes, but with current TrunkPort model, pulling it from agent would also pull the subports | 07:38 |
| kevinbenton | ihrachys: then we need to provide a way for the agent to query for the subports that belong to a trunk | 07:38 |
| armax | ihrachys: ack | 07:38 |
| ajo | as it's a field (ListOfObjectsField) I think | 07:38 |
| ihrachys | kevinbenton: why? you just listen to subports that have the trunk id that you want | 07:38 |
| kevinbenton | ihrachys: what do you think the agent should do when it starts? | 07:38 |
| armax | ajo: yes, but that happens when the agent knows nothing about the trunk | 07:38 |
| ajo | armax, yup | 07:39 |
| *** wolverineav has quit IRC | 07:39 | |
| kevinbenton | ihrachys: just list all subports? | 07:39 |
| ajo | armax, btw, I can get a real value for the RPC size on a 4k trunk port | 07:39 |
| *** bvandewa has joined #openstack-neutron | 07:39 | |
| ajo | if that's helpful | 07:39 |
| ihrachys | kevinbenton: ah right, for initial fetch, there indeed would need to be a method. | 07:39 |
| ihrachys | kevinbenton: sorry, need to run. I will ping you later | 07:39 |
| *** ihrachys has quit IRC | 07:40 | |
| kevinbenton | ihrachys: yeah, i would be okay with it stashing all subports | 07:40 |
| *** bvandewa has quit IRC | 07:40 | |
| armax | kevinbenton: what do you mean? | 07:40 |
| ajo | yeah, what do you mean :) | 07:40 |
| kevinbenton | armax: all agents keeping track of subport notifications | 07:40 |
| kevinbenton | armax: even if it's not currently assigned to them | 07:40 |
| armax | kevinbenton: well, I am trying to udnerstand how we can make the subport notification | 07:41 |
| armax | whether that’s stored or not, that’s a problem for later | 07:41 |
| ajo | yeah, that's true, one issue with this push API, is that currently, all agents will get it (fanout) | 07:41 |
| kevinbenton | armax: yeah, that's fine | 07:42 |
| kevinbenton | armax: ihar had just suggested killing the subports field on the trunk | 07:42 |
| armax | right now I think all agents get the push, just like you like it kevinbenton | 07:42 |
| ajo | you like that kevinbenton ? :P :) | 07:42 |
| kevinbenton | yes, the agents are like the borg. they should see everything | 07:42 |
| armax | kevinbenton: why? | 07:42 |
| *** jlibosva has joined #openstack-neutron | 07:42 | |
| kevinbenton | armax: to make the trunk payload small | 07:42 |
| ajo | we could eventually introduce a directed push | 07:42 |
| kevinbenton | ajo, armax: i like it because it helps for things that involve many agents | 07:43 |
| ajo | then the agent sees the trunk, it's for him, pulls a list of subports | 07:43 |
| kevinbenton | trunks it doesn't matter much because that's local to an agent | 07:43 |
| armax | kevinbenton: that would open up a can of worms | 07:43 |
| armax | kevinbenton: because to work, all trunks in the deployment need to be casted everywhere | 07:43 |
| armax | so that when a single VM boots up on an agent | 07:44 |
| armax | the agent knows what subports belong to the trunk to which the VM is attached to | 07:44 |
| kevinbenton | armax: yes, that's what the discussion was about initial retrieval | 07:44 |
| kevinbenton | armax: and why they would need to listen to all subports afterwards | 07:44 |
| ajo | yeah, we could change the pattern from 1) I get the Trunk port, will all the subports attached to | 07:45 |
| ajo | a) I get the Trunk port | 07:45 |
| armax | kevinbenton: it seems to me that we’d be designing to overcome a limitiation in the rpc middleware | 07:45 |
| ajo | b) I pull a list of subports with a filter | 07:45 |
| armax | when a ‘simple’ bulky push would do it | 07:45 |
| *** wolverineav has joined #openstack-neutron | 07:46 | |
| armax | ajo: actually | 07:46 |
| ajo | but b is not implemented yet either (in the pull interface, there are no filters, just object ids) | 07:46 |
| armax | ajo: another option to avoid the bulky push | 07:46 |
| armax | is use a bulky pull | 07:46 |
| kevinbenton | no | 07:46 |
| armax | I mean that works, doesn’t it? | 07:46 |
| kevinbenton | stop perpetuating the pull! | 07:46 |
| kevinbenton | :) | 07:47 |
| ajo | yeah, that's what I was proposing | 07:47 |
| ajo | lol | 07:47 |
| armax | kevinbenton: dude, give me a bulky push then | 07:47 |
| armax | :) | 07:47 |
| ajo | armax, kevinbenton let me measure the real damage of the RPC payload | 07:47 |
| ajo | may be it's not "that bad" | 07:47 |
| kevinbenton | armax: just push each subport as it's created/deleted | 07:47 |
| ajo | what if it's 256kB max? | 07:47 |
| ajo | wouldn't be that worrysome.. | 07:47 |
| *** chlong has quit IRC | 07:47 | |
| armax | kevinbenton: latency is gonna suck | 07:47 |
| kevinbenton | armax: stop trying to work around the rpc middleware :) | 07:48 |
| armax | kevinbenton: ok you clearly don’t get it | 07:48 |
| armax | I am going to bed :) | 07:48 |
| ajo | armax, have some rest, I'll report back with my thoughts & investigation | 07:48 |
| * ajo jumps to review https://review.openstack.org/#/c/347662/ now | 07:49 | |
| openstackgerrit | bin proposed openstack/neutron: Add routed network scenario test cases https://review.openstack.org/347188 | 07:49 |
| armax | ajo: thanks, if you could look into the feasibility of the builk operation great, otherwise we’ll figure ways around it | 07:49 |
| jschwarz | kevinbenton, hello again | 07:49 |
| ajo | armax, ack | 07:49 |
| kevinbenton | jschwarz: hey | 07:49 |
| armax | thanks! | 07:49 |
| armax | nite | 07:49 |
| jschwarz | kevinbenton, https://github.com/openstack/neutron/blob/master/neutron/db/l3_agentschedulers_db.py#L337 | 07:49 |
| ajo | good night! :) | 07:49 |
| openstackgerrit | Nir Magnezi proposed openstack/neutron: Adds a default reload callback to ProcessManager https://review.openstack.org/345263 | 07:50 |
| jschwarz | kevinbenton, why isn't the "filter_allocating_and_missing_routers" call happening before the other functions? | 07:50 |
| *** armax has quit IRC | 07:50 | |
| nmagnezi | jlibosva, ^^ fixed the test and added another one for the disable method :) | 07:50 |
| jschwarz | kevinbenton, i.e. line 329 (before get_ha_sync_data_for_host or get_sync_data)? | 07:50 |
| kevinbenton | jschwarz: because the routers can change state at any point during all of that data collection | 07:50 |
| kevinbenton | jschwarz: if we call first, none may be in ALLOCATING | 07:51 |
| kevinbenton | jschwarz: then immediately one switches and starts tearing down stuff | 07:51 |
| *** emagana has joined #openstack-neutron | 07:51 | |
| jschwarz | kevinbenton, blah | 07:51 |
| kevinbenton | jschwarz: which will screw up the interface retrieval | 07:51 |
| openstackgerrit | Nir Magnezi proposed openstack/neutron: Adds a default reload callback to ProcessManager https://review.openstack.org/345263 | 07:51 |
| nmagnezi | jlibosva, minor fix ^^ , sorry. | 07:51 |
| *** andymaier has joined #openstack-neutron | 07:51 | |
| jschwarz | kevinbenton, so another solution to that issue is having those functions also block on the CaS check (if it happens to merge) | 07:52 |
| jschwarz | kevinbenton, either way we also have to restore https://review.openstack.org/#/c/284400/ | 07:53 |
| kevinbenton | jschwarz: how will that help? | 07:53 |
| kevinbenton | jschwarz: they start and there is no lock | 07:53 |
| kevinbenton | jschwarz: then something takes the lock while they are retrieving info | 07:53 |
| jschwarz | kevinbenton, the "then something takes the lock" part isn't accurate, because it's a CaS | 07:54 |
| *** abregman_ is now known as abregman_|mtg | 07:54 | |
| jschwarz | can't retrieve something that is already taken | 07:54 |
| kevinbenton | jschwarz: it doesn't matter | 07:54 |
| kevinbenton | jschwarz: the point is that they look | 07:54 |
| kevinbenton | jschwarz: and then start retrieving data | 07:54 |
| kevinbenton | jschwarz: unless you are suggesting the sync data changes the state of the router | 07:55 |
| kevinbenton | jschwarz: then it won't help | 07:55 |
| jschwarz | kevinbenton, why not? | 07:55 |
| kevinbenton | jschwarz: sync data sees router is ACTIVE | 07:55 |
| kevinbenton | jschwarz: so it starts retrieving interfaces | 07:55 |
| kevinbenton | jschwarz: concurrently it changes to ALLOCATING and tears stuff apart | 07:56 |
| *** emagana has quit IRC | 07:56 | |
| *** bvandewa has joined #openstack-neutron | 07:56 | |
| kevinbenton | jschwarz: which breaks interface retrieval | 07:56 |
| jschwarz | kevinbenton, ah, so the router might not have the interfaces because the resources haven't been allocated yet? | 07:56 |
| kevinbenton | jschwarz: or the thing is being deleted | 07:56 |
| jlibosva | nmagnezi: hmm, now I see things I didn't see before :) | 07:56 |
| jschwarz | kevinbenton, iirc we don't change to ALLOCATING on deletion | 07:56 |
| nmagnezi | jlibosva, :< | 07:56 |
| kevinbenton | jschwarz: oh, i'm thinking of update to HA from legacy maybe | 07:57 |
| jschwarz | ah ok | 07:57 |
| kevinbenton | jschwarz: whatever the other ALLOCATING case was | 07:57 |
| kevinbenton | jschwarz: either way, this process looks first, then does stuff | 07:57 |
| kevinbenton | jschwarz: so checking for something and then proceeding is always racey | 07:57 |
| jschwarz | kevinbenton, iirc there's only one ALLOCATING case: when switching a router to be an HA (either on creation or an explicit update) | 07:57 |
| jschwarz | but yeah, I think you're right | 07:57 |
| kevinbenton | jschwarz: that's two cases :) | 07:58 |
| kevinbenton | jschwarz: at least two different ways from an API perspective | 07:58 |
| jschwarz | hmmm | 07:58 |
| jschwarz | on the create_router, the router is created from the get-go as ALLOCATING | 07:59 |
| korzen_ | kevinbenton, sorry I was AFK, so yes, for QoS it does not handles fields filtering, for legacy resources (port/subnet/network) all the infrastructure will be reused. And for QoS and for trunk API we should introduce some filtering mechanism, maybe similar to legacy one... | 07:59 |
| jschwarz | so blocking sync on the CaS should work | 07:59 |
| jschwarz | kevinbenton, the update_router is racey as you said I think | 07:59 |
| *** rossella_s has quit IRC | 07:59 | |
| *** zzzeek has quit IRC | 08:00 | |
| *** saggi has joined #openstack-neutron | 08:00 | |
| *** hoangcx has joined #openstack-neutron | 08:00 | |
| *** rossella_s has joined #openstack-neutron | 08:00 | |
| *** zzzeek has joined #openstack-neutron | 08:00 | |
| jschwarz | kevinbenton, though, on update_router to HA we explicitly unschedule the router from all agents | 08:00 |
| jschwarz | kevinbenton, is that what you meant by "tear down"? | 08:01 |
| *** matrohon has joined #openstack-neutron | 08:01 | |
| kevinbenton | jschwarz: agents can call in before it's unscheduled | 08:01 |
| *** ekuris has quit IRC | 08:01 | |
| *** salv-orlando has quit IRC | 08:01 | |
| kevinbenton | korzen_: yeah, it seems we need another decorator or an adjustment to the dict creation one | 08:02 |
| jlibosva | nmagnezi: commented, tell me wat da ya think? | 08:02 |
| jschwarz | kevinbenton, but then again, before the router is unscheduled it is changed to ALLOCATING (l3_hamode_db.py, line 540) | 08:03 |
| jschwarz | kevinbenton, so I do think blocking both sync and update on the CaS should work | 08:03 |
| jschwarz | kevinbenton, what am I missing? | 08:03 |
| *** abregman_|mtg has quit IRC | 08:03 | |
| kevinbenton | jschwarz: so you want sync to change the state to ALLOCATING? | 08:05 |
| jschwarz | kevinbenton, yes | 08:05 |
| *** andymaier has quit IRC | 08:05 | |
| kevinbenton | jschwarz: that means a user will see their router state flap whenever the agent is doing work | 08:06 |
| kevinbenton | jschwarz: like associating a FIP triggers this, no? | 08:06 |
| jschwarz | kevinbenton, yes | 08:06 |
| jschwarz | kevinbenton, the other solution is merging Liu's code that you linked earlier (lock-free stuff) | 08:06 |
| *** hanchao has joined #openstack-neutron | 08:06 | |
| jschwarz | kevinbenton, since the sync logic doesn't allocate stuff | 08:07 |
| jschwarz | or does it? | 08:07 |
| kevinbenton | jschwarz: yes, i would prefer to just skip it | 08:07 |
| kevinbenton | jschwarz: if it's not mutating things, i don't think we should change state | 08:07 |
| kevinbenton | jschwarz: i think it would if it calls auto_schedule | 08:07 |
| kevinbenton | jschwarz: but not just the sync_data | 08:07 |
| jschwarz | kevinbenton, _process_sync_ha_data does allocate actually | 08:07 |
| kevinbenton | jschwarz: ah | 08:07 |
| kevinbenton | jschwarz: well whatever that allocates can be blocked on cas | 08:08 |
| jschwarz | so it's technically violating the ALLOCATING agreement | 08:08 |
| kevinbenton | jschwarz: but i think we shouldn't on SYNC | 08:08 |
| kevinbenton | jschwarz: not really | 08:08 |
| kevinbenton | jschwarz: we don't return routers in ALLOCATING to the agent | 08:08 |
| *** _Fdaisuke_ has joined #openstack-neutron | 08:09 | |
| jschwarz | kevinbenton, that doesn't change the fact that _process_sync_ha_data does allocate things, and that can interfere with stuff that does use the CaS | 08:09 |
| jschwarz | and locking that code means that the user will see the state flapping as you said before | 08:10 |
| kevinbenton | jschwarz: no, there are two separate parts | 08:10 |
| kevinbenton | jschwarz: there is regular sync | 08:10 |
| *** Fdaisuke_ has quit IRC | 08:10 | |
| kevinbenton | jschwarz: which just retrieves data | 08:10 |
| kevinbenton | jschwarz: then you said there is a path that allocates things | 08:10 |
| kevinbenton | jschwarz: use ALLOCATING for that | 08:10 |
| kevinbenton | jschwarz: don't put ALLOCATING on the whole retrieval operation | 08:11 |
| *** obondarev has joined #openstack-neutron | 08:11 | |
| *** ushkalim has joined #openstack-neutron | 08:11 | |
| jschwarz | kevinbenton, sounds reasonable | 08:11 |
| jschwarz | kevinbenton, but then it will mean that on every sync_routers call from the l3 agent, the status of the router will flap | 08:11 |
| *** reedip has quit IRC | 08:11 | |
| kevinbenton | jschwarz: no | 08:11 |
| *** jschwarz has left #openstack-neutron | 08:11 | |
| kevinbenton | jschwarz: because it won't make changes | 08:11 |
| *** jschwarz has joined #openstack-neutron | 08:11 | |
| jschwarz | sorry, wrong key :P | 08:12 |
| jschwarz | kevinbenton, I think we might be talking about 2 different flows | 08:12 |
| kevinbenton | jschwarz: what is it that you are saying it does during sync_routers that allocates? | 08:12 |
| jschwarz | kevinbenton, sync_routesr -> list_active_sync_routers_on_active_l3_agent -> _get_active_l3_agent_routers_sync_data -> get_ha_sync_data_for_host -> _process_sync_ha_data | 08:14 |
| jschwarz | kevinbenton, that is basically an unconditional flow if I understand it correctly | 08:14 |
| kevinbenton | jschwarz: yes, and I don't see it allocating anything | 08:14 |
| jschwarz | oh wait | 08:14 |
| jschwarz | I think I confused something | 08:15 |
| *** ekuris has joined #openstack-neutron | 08:16 | |
| jschwarz | kevinbenton, yeah you're right | 08:16 |
| jschwarz | kevinbenton, sorry about that :) | 08:16 |
| *** chandankumar has quit IRC | 08:16 | |
| jschwarz | kevinbenton, so I agree, we don't want to get the ALLOCATING CaS in this flow at all and liuyulong's proposal is better | 08:17 |
| kevinbenton | jschwarz: unfortunately with the approach in that patch it's detecting a very specific failure related to the interface disappearing | 08:17 |
| *** roeyc has joined #openstack-neutron | 08:18 | |
| jschwarz | kevinbenton, yes | 08:18 |
| jschwarz | kevinbenton, there were a series of about 7? patches that dealt with the same thing a few months back as you might recall | 08:18 |
| jschwarz | kevinbenton, on the upside there aren't a lot of these cases left ;-) | 08:19 |
| *** lujinluo has quit IRC | 08:19 | |
| kevinbenton | jschwarz: yeah, i'm thinking of a way to generalize this | 08:20 |
| *** lucas-dinner is now known as lucasagomes | 08:21 | |
| *** vthapar has quit IRC | 08:21 | |
| kevinbenton | jschwarz: catch any errors occured during sync data retrieval for a router, then check to see if the router is gone or ALLOCATING | 08:21 |
| kevinbenton | jschwarz: if so, just ignore that router | 08:21 |
| jschwarz | kevinbenton, the thing is, we're looking for a solution that doesn't provide a state that is linearizable, but is as up to date as possible | 08:21 |
| jschwarz | kevinbenton, isn't that already what we are doing? | 08:22 |
| kevinbenton | jschwarz: no | 08:22 |
| kevinbenton | jschwarz: take that patch for example | 08:22 |
| kevinbenton | jschwarz: it's catching a very specific port not found error | 08:22 |
| kevinbenton | jschwarz: at a particular place | 08:22 |
| *** lujinluo has joined #openstack-neutron | 08:22 | |
| nmagnezi | jlibosva, i reviewed you comments | 08:23 |
| nmagnezi | jlibosva, i have questions about 3 of them | 08:23 |
| nmagnezi | jlibosva, lines 186, 188 , 190 | 08:23 |
| jschwarz | kevinbenton, I'm not sure I fully understand - are you thinking about throwing these exceptions and when catching them checking if the router has reverted to ALLOCATING concurrently? | 08:23 |
| kevinbenton | jschwarz: i'm suggesting "try: all_sync_retrieval_stuff, except: check_router_gone_or_allocating" | 08:24 |
| jschwarz | kevinbenton, and if it hasn't gone_or_allocating concurrently? | 08:24 |
| kevinbenton | jschwarz: then reraise because it's something else | 08:24 |
| jschwarz | then you have an actual error that you'll return to the API/agent? | 08:24 |
| jschwarz | hm | 08:24 |
| jschwarz | kevinbenton, a router might not be fully gone though | 08:25 |
| nmagnezi | jlibosva, starting with 186 , i'm not sure i understand why using Mock object, i provide the reload_callback when I initialize the manager anyways so that function becomes part of it, no? | 08:25 |
| *** edand has joined #openstack-neutron | 08:25 | |
| jschwarz | kevinbenton, for example, a router is in the process of being deleted so its HA resources have been deleted concurrently, but the router is not allocating and isn't completely gone | 08:25 |
| jschwarz | kevinbenton, solvable by switching a router to ALLOCATING when deleting it, though | 08:25 |
| jschwarz | kevinbenton, that's a nice approach. | 08:25 |
| kevinbenton | jschwarz: yeah, i think we had that before | 08:25 |
| jlibosva | nmagnezi: comments 186 says - this line is correct :) | 08:25 |
| jlibosva | comment* | 08:26 |
| kevinbenton | jschwarz: i can't remember why we decided against it | 08:26 |
| kevinbenton | jschwarz: in the original allocating patch | 08:26 |
| *** permalac has quit IRC | 08:26 | |
| kevinbenton | jschwarz: maybe it's placement didn't fit well | 08:26 |
| *** abregman has joined #openstack-neutron | 08:26 | |
| nmagnezi | jlibosva, oh, maybe i misread it. that is in oppose to lines 188 and 190 i presume | 08:26 |
| jschwarz | kevinbenton, things have changed since and we are all the wiser | 08:26 |
| jschwarz | kevinbenton, lets give it another go | 08:26 |
| *** permalac has joined #openstack-neutron | 08:26 | |
| *** ygbo has joined #openstack-neutron | 08:27 | |
| jlibosva | nmagnezi: L188 and L190 are just about that you don't explicitly need to create a mock object to an attribute of another mock object. as all attributes of basic mock objects are mock objects | 08:27 |
| *** reedip has joined #openstack-neutron | 08:27 | |
| nmagnezi | jlibosva, but it does not work the way you suggest: https://paste.fedoraproject.org/396685/46969446/ | 08:28 |
| jschwarz | kevinbenton, can you link me to that previous attempt you were talking about? | 08:28 |
| kevinbenton | jschwarz: let me see if i can find it | 08:29 |
| openstackgerrit | ChangBo Guo(gcb) proposed openstack/neutron: Use method get_ipv6_addr_by_EUI64 from oslo.utils https://review.openstack.org/347628 | 08:30 |
| jlibosva | nmagnezi: hmm, so maybe I talk rubbish. let me try it | 08:30 |
| openstackgerrit | Merged openstack/neutron: Add subresources support for PECAN https://review.openstack.org/336188 | 08:30 |
| *** tyrola has joined #openstack-neutron | 08:30 | |
| *** liuyulong has quit IRC | 08:33 | |
| *** liuyulong has joined #openstack-neutron | 08:34 | |
| *** numans has joined #openstack-neutron | 08:34 | |
| *** otherwiseguy has quit IRC | 08:34 | |
| jschwarz | kevinbenton, can you please mark this as Medium importance? https://bugs.launchpad.net/neutron/+bug/1533457 | 08:37 |
| openstack | Launchpad bug 1533457 in neutron "Neutron server unable to sync HA info after race between HA router creating and deleting" [Undecided,In progress] - Assigned to LIU Yulong (dragon889) | 08:37 |
| kevinbenton | jschwarz: yeah | 08:37 |
| *** roeyc has quit IRC | 08:37 | |
| *** roeyc1 has joined #openstack-neutron | 08:38 | |
| kevinbenton | jschwarz: not having luck finding it | 08:38 |
| kevinbenton | jschwarz: maybe it was in the context of the patch to filter ALLOCATING routers | 08:38 |
| jschwarz | kevinbenton, no worries | 08:38 |
| *** mhickey has joined #openstack-neutron | 08:39 | |
| *** manikanta_tadi has quit IRC | 08:39 | |
| *** bvandewa has quit IRC | 08:39 | |
| jschwarz | kevinbenton, a clean slate is the more adventurous way anyway ;-) | 08:40 |
| *** iranzo has joined #openstack-neutron | 08:40 | |
| *** iranzo has joined #openstack-neutron | 08:40 | |
| kevinbenton | jschwarz: yeah :) | 08:41 |
| kevinbenton | jschwarz: i think basically an update_status call at the top of delete_router :) | 08:42 |
| *** imcsk8 is now known as imcsk8|zZz | 08:42 | |
| *** otherwiseguy has joined #openstack-neutron | 08:43 | |
| *** zhhuabj has quit IRC | 08:43 | |
| jschwarz | kevinbenton, in addition to the "try-except" block we discussed earlier? that's what I had in mind as well :) | 08:43 |
| *** ramishra has quit IRC | 08:43 | |
| *** ramishra has joined #openstack-neutron | 08:44 | |
| kevinbenton | jschwarz: https://review.openstack.org/#/c/257059/15/neutron/db/l3_hamode_db.py | 08:45 |
| kevinbenton | jschwarz: now i need to figure out why it's gone :) | 08:45 |
| kevinbenton | jschwarz: removed in the very next revision | 08:45 |
| kevinbenton | jschwarz: with no explanation. thanks self :) | 08:45 |
| jschwarz | kevinbenton, lol | 08:45 |
| jschwarz | kevinbenton, so there are surely some IRC logs of us talking with probably amuller of why we didn't go through with this | 08:46 |
| *** zhhuabj has joined #openstack-neutron | 08:46 | |
| kevinbenton | jschwarz: either that or you are going to get some horrible failure explaining what we forgot to consider :) | 08:46 |
| *** mickeys has joined #openstack-neutron | 08:46 | |
| *** sambetts|afk is now known as sambetts | 08:46 | |
| *** saggi1 has joined #openstack-neutron | 08:47 | |
| *** saggi has quit IRC | 08:47 | |
| jschwarz | kevinbenton, http://eavesdrop.openstack.org/irclogs/%23openstack-neutron/%23openstack-neutron.2016-02-21.log.html#t2016-02-21T15:27:35 | 08:47 |
| jschwarz | not sure if that's for patchset 16 or 15 though | 08:48 |
| kevinbenton | jschwarz: ps15 was only up for like 45 mins before i clobbered it with 16 | 08:48 |
| jschwarz | kevinbenton, yeah, it might be 16 | 08:49 |
| kevinbenton | jschwarz: so give it a shot | 08:49 |
| jschwarz | kevinbenton, agreed | 08:49 |
| *** ekuris has quit IRC | 08:49 | |
| jschwarz | kevinbenton, just to clarify - when you said "give it a shot", did you mean "you should give it a shot" or "I should give it a shot"? | 08:50 |
| jschwarz | kevinbenton, a bit unclear if you're going to work on this or if I should ;-) | 08:50 |
| kevinbenton | jschwarz: no, you go ahead | 08:50 |
| jschwarz | kevinbenton, ack. be in touch :) | 08:51 |
| kevinbenton | jschwarz: unless you are busy | 08:51 |
| *** tonytan4ever has quit IRC | 08:51 | |
| jschwarz | kevinbenton, I'll fit it in among the other patches I need to write | 08:51 |
| *** zhhuabj_ has joined #openstack-neutron | 08:51 | |
| *** wolverineav has quit IRC | 08:51 | |
| kevinbenton | jschwarz: want to split it up? | 08:52 |
| kevinbenton | jschwarz: i could work on the exception catching, you can work on the state change on delete? | 08:52 |
| jschwarz | kevinbenton, sounds good | 08:52 |
| kevinbenton | jschwarz: ok cool | 08:52 |
| openstackgerrit | chandanc proposed openstack/neutron: The Iptables manager and firewall driver in Neutron must be enhanced for co-existence of SecurityGroup and FWaaS v2 APIs. This patch re-factors the IPTables driver for enabling FWaaS and SG chain to be interleaved preserving ordering of rules. https://review.openstack.org/348177 | 08:53 |
| *** wolverineav has joined #openstack-neutron | 08:53 | |
| *** andymaier has joined #openstack-neutron | 08:53 | |
| *** zhhuabj has quit IRC | 08:54 | |
| *** mickeys has quit IRC | 08:55 | |
| *** mickeys has joined #openstack-neutron | 08:55 | |
| openstackgerrit | Dongcan Ye proposed openstack/neutron: Remove validate for subports https://review.openstack.org/348181 | 08:56 |
| *** wolverineav has quit IRC | 08:57 | |
| *** ekuris has joined #openstack-neutron | 08:57 | |
| *** djan has quit IRC | 08:58 | |
| *** wolverineav has joined #openstack-neutron | 08:58 | |
| *** chandankumar has joined #openstack-neutron | 08:59 | |
| *** mfuruta has quit IRC | 09:00 | |
| *** mickeys has quit IRC | 09:00 | |
| *** zhhuabj has joined #openstack-neutron | 09:02 | |
| *** gouthamr has joined #openstack-neutron | 09:04 | |
| jschwarz | kevinbenton, so the issue with the delete part | 09:04 |
| kevinbenton | jschwarz: here it comes... :) | 09:04 |
| jschwarz | kevinbenton, is that the first thing that delete_router is hamode does, is to actually delete the router | 09:04 |
| jschwarz | all the resources come after | 09:04 |
| jschwarz | so it's not really needed :) | 09:04 |
| jschwarz | kevinbenton, https://review.openstack.org/#/c/257059/15/neutron/db/l3_hamode_db.py@535 | 09:04 |
| jschwarz | if the router is not gone by the time a race was discovered in the sync_routers, no resources have been deleted so there's no issue | 09:05 |
| *** zhhuabj_ has quit IRC | 09:05 | |
| *** stanzgy has quit IRC | 09:05 | |
| kevinbenton | jschwarz: well as part of delete_router, all interfaces that belong to it will be removed | 09:05 |
| jschwarz | kevinbenton, so only the "try-except: check_is_gone" portion is needed | 09:05 |
| kevinbenton | jschwarz: after my other change, the ha interface goes as part of super().delete_router :/ | 09:06 |
| jschwarz | kevinbenton, yeah, but the HA interfaces are deleted *after* the router is deleted | 09:06 |
| jschwarz | kevinbenton, the RouterPort change you mean? | 09:06 |
| kevinbenton | jschwarz: yep | 09:06 |
| jschwarz | ah | 09:06 |
| jschwarz | so.... line 554 is no longer needed | 09:06 |
| *** lixiaoy1 has joined #openstack-neutron | 09:07 | |
| kevinbenton | jschwarz: https://github.com/openstack/neutron/blob/374db46b9ec0ec5ed6697e85e8bbab4ca2de29b5/neutron/db/l3_hamode_db.py#L592-L597 | 09:07 |
| jschwarz | gah | 09:07 |
| *** zhhuabj has quit IRC | 09:07 | |
| *** javeriak has quit IRC | 09:07 | |
| kevinbenton | jschwarz: but it does mean you will now need the ALLOCATING thing | 09:07 |
| jschwarz | kevinbenton, anyway, the RouterPort change ensures the atomicity of the deletion, no? | 09:07 |
| *** javeriak has joined #openstack-neutron | 09:08 | |
| kevinbenton | jschwarz: no, it just means that the normal l3 stuff will call delete_port for us | 09:08 |
| jschwarz | kevinbenton, i.e. the ha interfaces are deleted as part of the same transaction as the deletion of a router. no? | 09:08 |
| jschwarz | ah ok | 09:08 |
| kevinbenton | jschwarz: nah, unfortunately we can't delete ports in a transaction with ML2 | 09:08 |
| *** wolverineav has quit IRC | 09:08 | |
| jschwarz | so it is needed now, but it also explains why it was removed before | 09:08 |
| jschwarz | :) | 09:08 |
| kevinbenton | jschwarz: yeah, probably :) | 09:08 |
| jschwarz | XD | 09:08 |
| openstackgerrit | Songming Yan proposed openstack/python-neutronclient: Add trunk commands to openstackclient https://review.openstack.org/340624 | 09:11 |
| jschwarz | kevinbenton, so I'm depending on https://review.openstack.org/#/c/343936/ | 09:11 |
| jschwarz | or is that not really needed? | 09:12 |
| openstackgerrit | MingShuang Xian proposed openstack/neutron-vpnaas: [WIP] Add external gateway port for VPN service https://review.openstack.org/341393 | 09:12 |
| jschwarz | I think it's not really related actually so never mind | 09:13 |
| *** claudiub|2 has joined #openstack-neutron | 09:13 | |
| *** wolverineav has joined #openstack-neutron | 09:16 | |
| openstackgerrit | Pawel Koniszewski proposed openstack/neutron: Test LM patch https://review.openstack.org/348187 | 09:18 |
| kevinbenton | jschwarz: yeah, should be orthoganal | 09:18 |
| jschwarz | kevinbenton, running a rally test now to see if it's at least not screwing up anything, then I' | 09:19 |
| jschwarz | I'll push it | 09:19 |
| *** lujinluo has quit IRC | 09:20 | |
| *** wolverineav has quit IRC | 09:20 | |
| *** mhickey has quit IRC | 09:20 | |
| *** mhickey has joined #openstack-neutron | 09:21 | |
| *** rahuls has quit IRC | 09:22 | |
| *** rahuls has joined #openstack-neutron | 09:22 | |
| *** gvrangan has joined #openstack-neutron | 09:22 | |
| kevinbenton | jschwarz: sgtm | 09:23 |
| *** permalac has quit IRC | 09:25 | |
| *** liuyulong has quit IRC | 09:26 | |
| *** rahuls_ has joined #openstack-neutron | 09:28 | |
| *** rahuls has quit IRC | 09:28 | |
| *** reedip has quit IRC | 09:29 | |
| *** javeriak has quit IRC | 09:29 | |
| *** obondarev has quit IRC | 09:31 | |
| *** lujinluo has joined #openstack-neutron | 09:32 | |
| *** liuyulong has joined #openstack-neutron | 09:33 | |
| jlibosva | nmagnezi: replied on review about the __get__.return_value failure | 09:34 |
| jlibosva | nmagnezi: I left you a syntax error in the code snippet I left there, it's an easter egg and it was definitely left there intentionally | 09:35 |
| *** obondarev has joined #openstack-neutron | 09:36 | |
| *** iwamoto has quit IRC | 09:36 | |
| *** wolverineav has joined #openstack-neutron | 09:39 | |
| jschwarz | kevinbenton, got: http://paste.openstack.org/show/543222/ | 09:41 |
| jschwarz | unsure if it's related so I'm removing the patch and re-running rally | 09:41 |
| *** hanchao has quit IRC | 09:43 | |
| *** yamamoto_ has quit IRC | 09:43 | |
| *** wolverineav has quit IRC | 09:44 | |
| *** claudiub|2 has quit IRC | 09:45 | |
| jschwarz | this reproduces without, but now it's happening for all the routers | 09:45 |
| jschwarz | kevinbenton, think I've hit yet another weird state :\ | 09:45 |
| nmagnezi | jlibosva, LOL | 09:47 |
| nmagnezi | jlibosva, looking now | 09:47 |
| kevinbenton | jschwarz: where's the rest of that traceback. is that it? | 09:47 |
| jschwarz | kevinbenton, yes | 09:47 |
| kevinbenton | jschwarz: is it actually causing failures? | 09:48 |
| jschwarz | kevinbenton, yes | 09:48 |
| jschwarz | kevinbenton, it actually caused every single create and delete in the second rally run to fail | 09:48 |
| kevinbenton | jschwarz: what was req-1452cfa0-1f25-4805-a99c-daac46b8c662 trying to do? | 09:48 |
| jschwarz | 2016-07-28 12:36:46.249 DEBUG neutron.plugins.ml2.plugin [req-1452cfa0-1f25-4805-a99c-daac46b8c662 c_rally_27be0b05_rFSiFp1C 4f3bee48989b4610aaccc86332b719f7] Deleting port 75a2329a-85a2-4358-aff4-3ac2ef078765 from (pid=7344) _pre_delete_port /opt/openstack/neutron/neutron/plugins/ml2/plugin.py:1500 | 09:49 |
| jschwarz | 2016-07-28 12:36:46.251 DEBUG neutron.callbacks.manager [req-1452cfa0-1f25-4805-a99c-daac46b8c662 c_rally_27be0b05_rFSiFp1C 4f3bee48989b4610aaccc86332b719f7] Notify callbacks for port, before_delete from (pid=7344) _notify_loop /opt/openstack/neutron/neutron/callbacks/manager.py:140 | 09:49 |
| jschwarz | 2016-07-28 12:36:46.252 DEBUG neutron.callbacks.manager [req-1452cfa0-1f25-4805-a99c-daac46b8c662 c_rally_27be0b05_rFSiFp1C 4f3bee48989b4610aaccc86332b719f7] Calling callback neutron.db.l3_db._prevent_l3_port_delete_callback from (pid=7344) _notify_loop /opt/openstack/neutron/neutron/callbacks/manager.py:147 | 09:49 |
| jschwarz | ah, the second failure is a PEBKAC | 09:50 |
| *** lujinluo has quit IRC | 09:50 | |
| *** manikanta_tadi has joined #openstack-neutron | 09:51 | |
| *** wolverineav has joined #openstack-neutron | 09:51 | |
| liuyulong | jschwarz, ping | 09:51 |
| *** jschwarz is now known as jschwarz|afk | 09:51 | |
| jschwarz|afk | gone for lunch, back in 20 minutes kevinbenton, liuyulong :) | 09:52 |
| kevinbenton | jschwarz|afk: ack | 09:52 |
| liuyulong | jschwarz, fine, the trace http://paste.openstack.org/show/543222/ seems related to this https://bugs.launchpad.net/neutron/+bug/1580899 | 09:53 |
| openstack | Launchpad bug 1580899 in neutron "Overlapped new router interface cannot remove" [Medium,In progress] - Assigned to LIU Yulong (dragon889) | 09:53 |
| *** ihrachys has joined #openstack-neutron | 09:54 | |
| liuyulong | and bug/1580899 is alternatively solved by this https://review.openstack.org/#/c/331732/ | 09:54 |
| *** saggi1 has quit IRC | 09:54 | |
| *** iyamahat has joined #openstack-neutron | 09:55 | |
| *** wolverineav has quit IRC | 09:55 | |
| *** lujinluo has joined #openstack-neutron | 09:57 | |
| openstackgerrit | Nir Magnezi proposed openstack/neutron: Adds a default reload callback to ProcessManager https://review.openstack.org/345263 | 09:57 |
| nmagnezi | jlibosva, ^^ all done. thanks a lot. | 09:58 |
| *** yamahata has quit IRC | 09:58 | |
| *** iyamahat has quit IRC | 10:01 | |
| *** ociuhandu has joined #openstack-neutron | 10:01 | |
| ajo | nmagnezi, looking too ;D | 10:01 |
| nmagnezi | ajo, thanks :-) | 10:01 |
| *** Alex_Stef has quit IRC | 10:02 | |
| ajo | oh nice | 10:02 |
| ajo | you reuse the logic of disable | 10:02 |
| ajo | good catch | 10:02 |
| nmagnezi | credit goes to jlibosva | 10:02 |
| nmagnezi | :) | 10:02 |
| *** wolverineav has joined #openstack-neutron | 10:03 | |
| korzen_ | ihrachys, are we allowing to filter subnet by admin_state_up, which is not an subnet field but network attribute, like http://127.0.0.1:9696/v2.0/subnets?admin_state_up=True in http://logs.openstack.org/01/321001/11/check/gate-neutron-dsvm-api/98d056e/console.html#_2016-07-27_14_47_10_906641 | 10:07 |
| ihrachys | korzen_: apparently yes :) | 10:08 |
| *** salv-orlando has joined #openstack-neutron | 10:08 | |
| *** hoangcx has quit IRC | 10:08 | |
| *** devvesa has quit IRC | 10:08 | |
| korzen_ | ihrachys, :( how so? | 10:09 |
| ihrachys | korzen_: apparently there is some query filter hook registered? | 10:09 |
| korzen_ | ihrachys, yea, need to find it ;) | 10:10 |
| openstackgerrit | Davanum Srinivas (dims) proposed openstack/neutron: [WIP] Testing latest u-c https://review.openstack.org/303346 | 10:10 |
| *** wolverineav has quit IRC | 10:10 | |
| openstackgerrit | Kevin Benton proposed openstack/neutron: Protect _get_active_l3..._data from update/delete https://review.openstack.org/348215 | 10:11 |
| *** itzikb has quit IRC | 10:13 | |
| *** wolverineav has joined #openstack-neutron | 10:14 | |
| ihrachys | korzen_: but 1 failed is promising :) | 10:14 |
| jschwarz|afk | kevinbenton, liuyulong, back | 10:15 |
| jschwarz|afk | kevinbenton, can i interest you with more logs? :) | 10:15 |
| *** jschwarz|afk is now known as jschwarz | 10:15 | |
| *** salv-orlando has quit IRC | 10:15 | |
| kevinbenton | jschwarz: it looks like rally is trying to delete the port directly | 10:15 |
| kevinbenton | jschwarz: is that really the first appearance of that request ID? | 10:16 |
| jschwarz | kevinbenton, yes | 10:16 |
| kevinbenton | jschwarz: usually there is something from the API layer | 10:16 |
| kevinbenton | jschwarz: i wonder where it came from | 10:16 |
| jschwarz | let me have a look again | 10:16 |
| *** Alex_Stef has joined #openstack-neutron | 10:16 | |
| jschwarz | kevinbenton, yeah, that was the first appearance | 10:17 |
| jschwarz | weird bastard | 10:17 |
| *** buttercup has quit IRC | 10:17 | |
| kevinbenton | jschwarz: does rally log it's requests? | 10:17 |
| kevinbenton | jschwarz: like tempest does? | 10:17 |
| kevinbenton | jschwarz: if you can find the request id you can see what triggered it | 10:18 |
| jschwarz | kevinbenton, afraid not | 10:18 |
| *** wolverineav has quit IRC | 10:19 | |
| *** boden has joined #openstack-neutron | 10:20 | |
| jschwarz | kevinbenton, so there are a few more lines with that request id later on | 10:20 |
| jschwarz | kevinbenton, specifically: INFO neutron.wsgi [req-1452cfa0-1f25-4805-a99c-daac46b8c662 c_rally_27be0b05_rFSiFp1C 4f3bee48989b4610aaccc86332b719f7] 10.35.6.16 - - [28/Jul/2016 12:36:46] "DELETE //v2.0/ports/75a2329a-85a2-4358-aff4-3ac2ef078765.json HTTP/1.1" 409 423 0.624804 | 10:20 |
| jschwarz | so it does seem like a port deletion by rally | 10:20 |
| openstackgerrit | Merged openstack/neutron: Make create_object_with_dependency cleanup https://review.openstack.org/292950 | 10:22 |
| kevinbenton | jschwarz: rally cleanup stuff getting confused and trying to get rid of router ports? | 10:22 |
| jschwarz | kevinbenton, maybe. rally does create a few resources on its own, so maybe it's also trying to clean it up differently | 10:22 |
| liuyulong | kevinbenton, hi, https://review.openstack.org/#/c/265680/ does this patch need the change you just suggested? | 10:22 |
| *** liyifeng has joined #openstack-neutron | 10:23 | |
| kevinbenton | liuyulong: yes, did you see my response? | 10:23 |
| kevinbenton | liuyulong: this will avoid an extra call to the core plugin with the extension processing that comes with it | 10:24 |
| liuyulong | jschwarz, this patch, https://review.openstack.org/#/c/265672/, seems the L3 dvr_snat/legacy agent still have chance to get a HA router without ha_port. | 10:24 |
| jschwarz | liuyulong, yes, I saw your comment about this on that patch | 10:25 |
| jschwarz | liuyulong, I'll do my best to have a look at the code later on and reply there :) | 10:25 |
| liuyulong | kevinbenton, I've test such change before, the race may happen in that binding.port check line. | 10:25 |
| kevinbenton | liuyulong: how? | 10:25 |
| kevinbenton | liuyulong: that triggers a relationship lookup by sqlalchemy | 10:26 |
| kevinbenton | liuyulong: if it's deleted, it returns none | 10:26 |
| kevinbenton | liuyulong: if it's not, you'll get a port | 10:26 |
| *** devvesa has joined #openstack-neutron | 10:26 | |
| liuyulong | kevinbenton, patch set 12, I've done that change. | 10:26 |
| kevinbenton | liuyulong: and what happened? | 10:28 |
| jschwarz | kevinbenton, there's a rally's delete port action that's going on and is failing | 10:28 |
| jschwarz | so you were right | 10:28 |
| kevinbenton | jschwarz: i wonder why that started | 10:29 |
| liuyulong | jschwarz, thanks, that `if not ha_port:` check need remove someday. | 10:29 |
| kevinbenton | jschwarz: rally change? | 10:29 |
| jschwarz | kevinbenton, no idea... my setup is not responding :\ | 10:29 |
| liuyulong | kevinbenton, rally tests shows a None type error. | 10:29 |
| liuyulong | akamyshnikova_, ping? | 10:30 |
| jschwarz | liuyulong, Ann is on PTO | 10:30 |
| kevinbenton | liuyulong: do you have a traceback? | 10:31 |
| korzen_ | ihrachys, it does not seems to work, the admin_state_up I can set to True or False and get the same result... | 10:31 |
| korzen_ | ihrachys, I see that network and port have admin_state_up and test checks for port/subnet/network, so passing admin_state_up to every resource... | 10:33 |
| *** john-davidge has joined #openstack-neutron | 10:33 | |
| kevinbenton | liuyulong: i would rather we fix that approach | 10:33 |
| kevinbenton | liuyulong: drop the joined load | 10:33 |
| liuyulong | kevinbenton, it's same to the LP bug paste, not really has that patch's trace. | 10:33 |
| kevinbenton | liuyulong: because that impacts other things like delete that don't use the relationhip | 10:33 |
| ihrachys | korzen_: yeah, so probably it swallowed it silently before | 10:33 |
| ihrachys | korzen_: I guess we should modify the test | 10:33 |
| *** wolverineav has joined #openstack-neutron | 10:34 | |
| *** kawa2014 has quit IRC | 10:34 | |
| ihrachys | I think it's legit to return an error | 10:34 |
| ihrachys | thoughts? | 10:34 |
| *** john-dav_ has joined #openstack-neutron | 10:34 | |
| liuyulong | jschwarz, Ann is not here, so do you have a rally envrionment? | 10:34 |
| korzen_ | ihrachys, hmm error or drop silently not known filters | 10:34 |
| liuyulong | kevinbenton, OK, I'll change that. | 10:35 |
| *** buttercup has joined #openstack-neutron | 10:35 | |
| ihrachys | korzen_: I find it valid if we validate (fail on) filters | 10:35 |
| *** iranzo_ has joined #openstack-neutron | 10:35 | |
| ihrachys | if we don't know how to fulfill a request, it's better to crash than return random data | 10:35 |
| jschwarz | liuyulong, I do | 10:35 |
| ihrachys | but that's probably something to discuss in the team | 10:35 |
| ihrachys | maybe an email to opentack-dev@? | 10:35 |
| kevinbenton | liuyulong: left one more comment | 10:35 |
| kevinbenton | liuyulong: assign binding.port to a var | 10:36 |
| ihrachys | just so everyone is aware we enforce strict validation for filters. | 10:36 |
| kevinbenton | liuyulong: in case sqlalchemy is somehow expiring it | 10:36 |
| korzen_ | ihrachys, in my opinion is also valid to error out... ok I will modify the test for now and drop question on ML | 10:36 |
| ihrachys | korzen_: cool | 10:36 |
| kevinbenton | liuyulong: unless it's a weakref we shouldn't get Nonetype errors | 10:36 |
| *** emagana has joined #openstack-neutron | 10:36 | |
| ihrachys | jlibosva: re circular thing, nope, pep8 still passes locally | 10:36 |
| openstackgerrit | John Schwarz proposed openstack/neutron: Mark an HA router as ALLOCATING before deletion https://review.openstack.org/348224 | 10:37 |
| jschwarz | kevinbenton, ^ | 10:37 |
| *** iranzo has quit IRC | 10:37 | |
| *** iranzo_ is now known as iranzo | 10:37 | |
| jlibosva | ihrachys: even with tox -repep8 ? to rebuild the env? | 10:37 |
| *** john-davidge has quit IRC | 10:37 | |
| liuyulong | jschwarz, I will submit a new patch very soon, https://review.openstack.org/#/c/266109/ here, could you please run some rally test? | 10:37 |
| jschwarz | liuyulong, I could later on | 10:37 |
| jschwarz | liuyulong, can you tell me though exactly what I'll be looking for? | 10:38 |
| jschwarz | kevinbenton, re: those weird ports | 10:38 |
| jschwarz | kevinbenton, the port's device_owner is network:ha_router_replicated_interface | 10:38 |
| *** sdague has joined #openstack-neutron | 10:39 | |
| liuyulong | jschwarz, sorry, wrong patch, it's https://review.openstack.org/265680 | 10:40 |
| kevinbenton | jschwarz: why is it trying to delete them? | 10:40 |
| jschwarz | kevinbenton, some logic of rally | 10:40 |
| jschwarz | I think my neutron-server has entered some endless loop because the machine is taking a long time to responde | 10:40 |
| jschwarz | liuyulong, so i'm looking for the "id": port["id"] error? | 10:41 |
| *** emagana has quit IRC | 10:41 | |
| liuyulong | jschwarz, yes, and see if L3 agent have some ha_port check error. | 10:42 |
| jschwarz | liuyulong, can you provide an exact traceback, so that there are no mistakes? | 10:42 |
| *** wolverineav has quit IRC | 10:42 | |
| *** tbachman_ has joined #openstack-neutron | 10:43 | |
| jschwarz | kevinbenton, this is the rally code I ran: https://github.com/openstack/rally/blob/master/rally/plugins/openstack/scenarios/neutron/network.py#L214 | 10:44 |
| *** tbachman has quit IRC | 10:45 | |
| *** tbachman_ is now known as tbachman | 10:45 | |
| *** yamamoto has joined #openstack-neutron | 10:46 | |
| kevinbenton | jschwarz: where is the port delete coming from? | 10:46 |
| *** manikanta_tadi is now known as manikanta_afk | 10:46 | |
| *** manikanta_afk is now known as manikanta_tadi | 10:46 | |
| *** itzikb has joined #openstack-neutron | 10:47 | |
| liuyulong | jschwarz, http://paste.openstack.org/show/473839/ (neutron server), http://paste.openstack.org/show/523757/ (l3 agent) and http://paste.openstack.org/show/528407/ (l3 agent infinite loop) | 10:47 |
| jschwarz | kevinbenton, I'm not sure | 10:49 |
| jschwarz | kevinbenton, I'll try to dig into it a bit more and get back to you if I find something | 10:49 |
| *** yamamoto has quit IRC | 10:49 | |
| *** chandankumar has quit IRC | 10:50 | |
| kevinbenton | jschwarz: ack | 10:51 |
| *** chandankumar has joined #openstack-neutron | 10:51 | |
| openstackgerrit | Ihar Hrachyshka proposed openstack/neutron: Introduce ovo objects for security groups https://review.openstack.org/284738 | 10:51 |
| *** wolverineav has joined #openstack-neutron | 10:52 | |
| jschwarz | kevinbenton, also, https://review.openstack.org/#/c/348215/ should probably depend on https://review.openstack.org/#/c/348224/ | 10:52 |
| ihrachys | korzen_: fyi we need https://review.openstack.org/#/c/284738/48/neutron/objects/base.py to be able to define custom filters on a class | 10:52 |
| ihrachys | in case you'll need it | 10:52 |
| ihrachys | I should split it from the patch | 10:52 |
| openstackgerrit | Kevin Benton proposed openstack/neutron: Protect _get_active_l3..._data from update/delete https://review.openstack.org/348215 | 10:53 |
| kevinbenton | jschwarz: done | 10:53 |
| *** yamamoto has joined #openstack-neutron | 10:54 | |
| *** thorst has joined #openstack-neutron | 10:56 | |
| korzen_ | ihrachys, what exactly you have in mind? | 10:57 |
| ihrachys | korzen_: I needed that for sec group to be able to search by is_default synthetic field: https://review.openstack.org/#/c/284738/48/neutron/objects/securitygroup.py | 10:57 |
| ihrachys | line 48 | 10:57 |
| korzen_ | ihrachys, jlibosva and I have added it in __init__ | 10:58 |
| korzen_ | ihrachys, it is not working for you? | 10:58 |
| ihrachys | korzen_: could be, but then it's per instance | 10:58 |
| ihrachys | I prefer a declarative approach. | 10:59 |
| ihrachys | is there an issue with the change? | 10:59 |
| * jlibosva catches up | 10:59 | |
| openstackgerrit | Takashi NATSUME proposed openstack/neutron-lib: Add a hacking rule for string interpolation at logging https://review.openstack.org/347126 | 10:59 |
| korzen_ | jlibosva, adding extra_filter_names | 10:59 |
| *** shausy has quit IRC | 11:00 | |
| korzen_ | ihrachys, I guess it would be consistent to do it your way | 11:00 |
| *** jhershbe has quit IRC | 11:00 | |
| korzen_ | ihrachys, than adding init | 11:00 |
| jlibosva | korzen_: ihrachys you mean setting it up in __int__ instead of metaclass? | 11:00 |
| jlibosva | __init__ | 11:00 |
| ihrachys | yeah, otherwise why do we describe the attribute in base class if you can't just set it | 11:00 |
| ihrachys | jlibosva: for declarative metaclass that's ok | 11:01 |
| ihrachys | jlibosva: I think about objects that want to add a filter or too | 11:01 |
| ihrachys | no custom metaclasses | 11:01 |
| ihrachys | just define the attribute on the class as we do for other stuff | 11:01 |
| *** wolverineav has quit IRC | 11:02 | |
| jlibosva | ok, let me try to remember. I think there was a reason to do it in metaclass | 11:02 |
| *** obondarev has quit IRC | 11:02 | |
| *** roeyc1 has quit IRC | 11:02 | |
| *** obondarev has joined #openstack-neutron | 11:04 | |
| jlibosva | ihrachys: so the plan is to define it on base class and in metaclass create a copy? | 11:04 |
| kevinbenton | korzen_: so _make_subnet_dict no longer has access to the db model? | 11:04 |
| jlibosva | ihrachys: otherwise, inherited subclasses will have the same set that could lead to adding filter hooks to superclass | 11:04 |
| openstackgerrit | LIU Yulong proposed openstack/neutron: Ensure HA router can be synced after HA router race conditions https://review.openstack.org/265680 | 11:05 |
| liuyulong | jschwarz, kevinbenton, ^, please review. | 11:05 |
| ihrachys | jlibosva: yes, I think we want to copy because it may be the very base set | 11:05 |
| ihrachys | kevinbenton: korzen_: btw re access to db model from object, I think we should be cautious about it BUT we could add a property on objects that would care corresponding model if fetched. | 11:06 |
| kevinbenton | ihrachys: well unless we intend to break out-of-tree extensions, we need to pass _make_whatever_dict the db model | 11:07 |
| *** ramishra has quit IRC | 11:08 | |
| ihrachys | kevinbenton: there is reason _make_whatever_dict is named in such a way... :) | 11:08 |
| korzen_ | kevinbenton, make_whatever_dict will operate on object not db model | 11:08 |
| kevinbenton | korzen_: that won't work | 11:08 |
| *** jlanoux has quit IRC | 11:08 | |
| kevinbenton | korzen_: if the extension references a relationship on the model | 11:09 |
| ihrachys | kevinbenton: in general case, object mostly mimics models. | 11:09 |
| ihrachys | kevinbenton: we have synthetic fields that pull in objects thru relationships, so it will also mostly work | 11:09 |
| *** Alex_Stef has quit IRC | 11:09 | |
| korzen_ | kevinbenton, when you define new extension, you have to add it as new field in object | 11:09 |
| kevinbenton | ihrachys: they auto iterate all relationships? | 11:09 |
| korzen_ | kevinbenton, you have to add it | 11:10 |
| kevinbenton | korzen_: how do i do that as an out of tree driver? | 11:10 |
| *** iranzo has quit IRC | 11:10 | |
| korzen_ | kevinbenton, you can add it like in out of tree driver code | 11:10 |
| kevinbenton | korzen_: can a driver/service plugin dynamically change which fields an object has? | 11:11 |
| ihrachys | kevinbenton: you define what you pull, but yeah, there is a high level of automation. f.e. sec group object provides 'rules' fields that is a list of objects that corresponds to model.rules attr | 11:11 |
| liuyulong | jschwarz, I'll appreciate for your testing. It's 7:00 p.m. in Beijing now, time to go home. : ) | 11:11 |
| korzen_ | kevinbenton, it is doable to add fields, but you have to bump the object version | 11:11 |
| kevinbenton | korzen_: and how do i do that without changing the neutron code? | 11:12 |
| *** ramishra has joined #openstack-neutron | 11:12 | |
| jschwarz | liuyulong, I had to re-stack so that will take a few minutes | 11:12 |
| jschwarz | liuyulong, and I have a meeting in 20 minutes so it might take a bit longer | 11:12 |
| jschwarz | liuyulong, I'll run the rally test and comment on the review with my results.. sounds good? | 11:12 |
| ihrachys | kevinbenton: aren't we going to control api definition? | 11:13 |
| liuyulong | jschwarz, yeah, thank you very much. | 11:13 |
| jschwarz | liuyulong, :) | 11:13 |
| kevinbenton | ihrachys: there was no agreement that we were going to break extensions was there? | 11:13 |
| kevinbenton | ihrachys: at the summit that suggestion did not land well | 11:14 |
| ihrachys | kevinbenton: no, I don't think in general we should break. that's why I say we can carry db model on the object and pass it into _make_whatever. | 11:14 |
| ihrachys | but I really think that we should have a path away from it. | 11:14 |
| korzen_ | ihrachys, kevinbenton and that is why versioned objects are not crafted for neutron, extension, extension everywhere... | 11:16 |
| *** shz has quit IRC | 11:16 | |
| korzen_ | kevinbenton, but you have a valid point | 11:16 |
| *** shz has joined #openstack-neutron | 11:16 | |
| kevinbenton | ihrachys: sure, but the community has to agree that extensions are being killed | 11:16 |
| kevinbenton | we can't silently break things | 11:17 |
| ihrachys | kevinbenton: sure. let's explore passing models into _make_*, it is easy to do | 11:17 |
| *** obondarev has quit IRC | 11:17 | |
| openstackgerrit | John Schwarz proposed openstack/neutron: Mark an HA router as ALLOCATING before deletion https://review.openstack.org/348224 | 11:17 |
| kevinbenton | ihrachys: perhaps we can pass them a wrapper to a db model | 11:17 |
| kevinbenton | ihrachys: that emits an ugly deprecation warning | 11:18 |
| kevinbenton | ihrachys: if they use a non-model column | 11:18 |
| ihrachys | kevinbenton: but then, what are they going to switch to to stop getting the warning? | 11:18 |
| kevinbenton | ihrachys: non-OVO* column | 11:18 |
| *** devvesa has quit IRC | 11:18 | |
| *** diga has quit IRC | 11:18 | |
| kevinbenton | ihrachys: propose their stuff upstream | 11:18 |
| ihrachys | kevinbenton: ah, agreed, that will do | 11:18 |
| kevinbenton | ihrachys: but even that should probably wait until start of O | 11:19 |
| kevinbenton | ihrachys: unless we get agreement right away to deprecate | 11:19 |
| kevinbenton | ihrachys: so yeah, db model access to _make_*_dict for now | 11:19 |
| ihrachys | kevinbenton: I think the first step should be just passing a model, then when we have better object adoption in tree, we consider deprecation technics | 11:19 |
| ihrachys | kevinbenton: korzen_: I will craft a small patch to expose models on objects right now | 11:20 |
| kevinbenton | ihrachys: +1 | 11:20 |
| korzen_ | ihrachys, ok, thanks | 11:20 |
| korzen_ | I'm on PTO on Friday | 11:20 |
| korzen_ | ihrachys, please update me in a mail | 11:21 |
| kevinbenton | ihrachys: i was also chatting with korzen_, the current pattern being followed for native OVO stuff (e.g. qos and trunk) leaks things which aren't actually supported yet by a service pluging | 11:21 |
| ihrachys | korzen_: I will add you to reviewers | 11:21 |
| korzen_ | ihrachys, and I'm leaving office in na hour | 11:21 |
| ihrachys | kevinbenton: elaborate on leaks | 11:21 |
| *** Alex_Stef has joined #openstack-neutron | 11:22 | |
| kevinbenton | ihrachys: revision_id is showing up in the API | 11:22 |
| openstackgerrit | John Schwarz proposed openstack/neutron: Mark an HA router as ALLOCATING before deletion https://review.openstack.org/348224 | 11:22 |
| kevinbenton | ihrachys: even though the extension/API definition for it hasn't merged | 11:22 |
| ihrachys | huh? | 11:22 |
| kevinbenton | ihrachys: for trunk | 11:22 |
| jschwarz | kevinbenton, ^ your wish is my command | 11:22 |
| kevinbenton | ihrachys: the create method just uses that _to_dict decorator | 11:22 |
| kevinbenton | ihrachys: which happily returns everything | 11:23 |
| kevinbenton | ihrachys: so it includes everything on the object | 11:23 |
| kevinbenton | ihrachys: which includes revision from my work | 11:23 |
| kevinbenton | ihrachys: however, that shouldn't be sent to the API level | 11:23 |
| ihrachys | kevinbenton: got it. probably same issue with other stdattributes? | 11:24 |
| ihrachys | description? | 11:24 |
| kevinbenton | ihrachys: yep | 11:24 |
| ihrachys | ok, lemme think | 11:24 |
| kevinbenton | ihrachys: well description and created_at already have service plugins | 11:24 |
| kevinbenton | ihrachys: so they are legitimate | 11:24 |
| kevinbenton | ihrachys: long story short we are missing a call to self.fields i believe | 11:24 |
| ihrachys | kevinbenton: you mean _fields? | 11:25 |
| kevinbenton | yeah | 11:25 |
| ihrachys | kevinbenton: I don't think it will change much, fields are not generally passed, and they are not for that matter I think | 11:26 |
| kevinbenton | ihrachys: oh yeah | 11:26 |
| kevinbenton | ihrachys: that's just additional filtering | 11:26 |
| ihrachys | they are passed by user to limit the output | 11:26 |
| ihrachys | yeah. so we need some way to claim some fields belonging to specific extensions, and filter based on that. | 11:26 |
| kevinbenton | ihrachys: yeah, we would basically need the to_dict to not put on things that belong to extensions | 11:27 |
| kevinbenton | ihrachys: if it's too annoying to fix in OVO we can probably kill these at the API layer | 11:28 |
| ihrachys | we have ext attrs maps defined already, but I don't know how fragile that could be if we try to filter based on that | 11:28 |
| *** zhhuabj has joined #openstack-neutron | 11:28 | |
| *** john-dav_ is now known as john-davidge | 11:29 | |
| kevinbenton | ihrachys: are those a list of things from extensions? | 11:29 |
| ihrachys | I think it belongs to API layer, since it better knows what's enabled and which new attr map definitions are used. | 11:29 |
| kevinbenton | ihrachys: oh, not something specific to ovo | 11:29 |
| kevinbenton | ihrachys: yeah, don't touch the api layer from ovo | 11:30 |
| *** devvesa has joined #openstack-neutron | 11:30 | |
| ihrachys | kevinbenton: I mean, capture RESOURCE_ATTRIBUTE_MAP when it's registered, with the relation to extension name registered; then filter those out unless a plugin provides support. | 11:30 |
| *** zioproto has joined #openstack-neutron | 11:30 | |
| kevinbenton | ihrachys: yeah, that's API logic | 11:30 |
| kevinbenton | ihrachys: i was thinking explicit list definition on the object | 11:31 |
| kevinbenton | ihrachys: 'fields_from_extensions' | 11:31 |
| ihrachys | yeah, but then you kinda duplicate definitions from neutron/extensions/ :) | 11:31 |
| zioproto | Hello is anyone using the LBaaS in Kilo ? LBaas V1 I guess. I am getting crazy it looks like I cannot use the same floating IP to load balance on the same floating IP both http and https | 11:31 |
| kevinbenton | ihrachys: then to_dict could take a param to exclude them if requested | 11:31 |
| ihrachys | so every time you need something, you do it in both places. | 11:31 |
| kevinbenton | ihrachys: they already have to do that because they have to register the field in the object | 11:32 |
| *** jlanoux has joined #openstack-neutron | 11:32 | |
| kevinbenton | ihrachys: it might be easier to list which fields don't come from extensions... :) | 11:32 |
| ihrachys | haha | 11:32 |
| *** salv-orlando has joined #openstack-neutron | 11:33 | |
| ihrachys | kevinbenton: I can look at that today, a bit later, after I cover the easier part which is db_model on objects | 11:33 |
| *** rtheis has joined #openstack-neutron | 11:34 | |
| ihrachys | kevinbenton: great feedback so far, thanks for that. | 11:34 |
| kevinbenton | ihrachys: ack. if you find the only way to fix is to start peeking into the resource attr stuff, just stop and we'll filter at the API layer | 11:34 |
| kevinbenton | ihrachys: which should already be doing this anyway | 11:34 |
| kevinbenton | ihrachys: it does it on input | 11:35 |
| kevinbenton | ihrachys: so why not output | 11:35 |
| openstackgerrit | Artur Korzeniewski proposed openstack/neutron: objects: Subnet OVO usage in db base plugin code for get, update and delete https://review.openstack.org/321001 | 11:35 |
| ihrachys | kevinbenton: well, korzen_ today noticed that API allows to pass thru random filters like admin_state_up for subnets, even though it does not work. | 11:35 |
| kevinbenton | ihrachys: yes, i already replied to that email | 11:36 |
| ihrachys | kevinbenton: so not sure either input is covered | 11:36 |
| kevinbenton | ihrachys: we can't change that | 11:36 |
| ihrachys | ok will chec | 11:36 |
| ihrachys | *check | 11:36 |
| kevinbenton | ihrachys: it's backwards incompatible and will break cache busting crap like "GET /v2.0/ports.json?name=myport&nonce=a987cd987d | 11:36 |
| *** gongysh has quit IRC | 11:37 | |
| *** ratailor has quit IRC | 11:37 | |
| ihrachys | I see. we probably want to have a strict mode, and then a forgiving one where we pass filters from API? | 11:38 |
| ihrachys | and then in v3 :) we'll be able to revisit | 11:38 |
| kevinbenton | ihrachys: i don't even think we want this configurable | 11:39 |
| kevinbenton | ihrachys: if that's what you mean | 11:39 |
| ihrachys | kevinbenton: I mean, on object API level | 11:39 |
| ihrachys | kevinbenton: let's say the default is strict, but then if you pass forgiving=True, it ignores | 11:40 |
| kevinbenton | ihrachys: sure, we can also pull non-matching ones out at the API level | 11:40 |
| ihrachys | in python, not API or config files | 11:40 |
| kevinbenton | ihrachys: API knows all valid fields | 11:41 |
| ihrachys | kevinbenton: then again, you need to know what's allowed in db, which may be tricky with query hooks and all | 11:41 |
| kevinbenton | ihrachys: API only allows you to query based on visible fields in the API though, right? | 11:41 |
| *** liuyulong has quit IRC | 11:42 | |
| ihrachys | kevinbenton: then why admin_state_up passed into objects? | 11:42 |
| *** ramishra has quit IRC | 11:42 | |
| ihrachys | for subnet | 11:42 |
| korzen_ | ihrachys, kevinbenton it seems like apply it for everything ;) | 11:44 |
| kevinbenton | https://github.com/openstack/neutron/blob/master/neutron/db/common_db_mixin.py#L203 | 11:44 |
| korzen_ | https://review.openstack.org/#/c/321001/13/neutron/tests/tempest/api/test_auto_allocated_topology.py@61 | 11:44 |
| *** gouthamr has quit IRC | 11:44 | |
| kevinbenton | yeah, just looking for an attribute on the model | 11:44 |
| kevinbenton | but we can throw away stuff based on the visible fields | 11:45 |
| *** karthiks has quit IRC | 11:45 | |
| *** jhershbe has joined #openstack-neutron | 11:45 | |
| *** banix has joined #openstack-neutron | 11:45 | |
| kevinbenton | i'm fine breaking queries if people managed to get them to work on hidden fields | 11:45 |
| korzen_ | kevinbenton, thanks for good feedback, I will reach you on Monday for extension handling and API filtering, I must be going now | 11:46 |
| kevinbenton | korzen_: k, ttyl | 11:46 |
| *** yuelongguang has quit IRC | 11:47 | |
| *** garyk1 has quit IRC | 11:48 | |
| *** saggi has joined #openstack-neutron | 11:48 | |
| *** gvrangan has quit IRC | 11:48 | |
| openstackgerrit | Ihar Hrachyshka proposed openstack/neutron: Introduce ovo objects for security groups https://review.openstack.org/284738 | 11:48 |
| kevinbenton | ihrachys: fun. neutron net-list --standard-attr-id=4 | 11:48 |
| kevinbenton | ihrachys: filters based on standard attr id, which they can never see :) | 11:49 |
| ihrachys | meh | 11:49 |
| *** boden has quit IRC | 11:49 | |
| ihrachys | that sucks | 11:49 |
| kevinbenton | ihrachys: through a painful enumeration process they could discover their standard attrs | 11:49 |
| kevinbenton | ihrachys: yeah, i think the API layer can dump this crap | 11:49 |
| kevinbenton | ihrachys: so OVO doesn't have to worry about ti | 11:49 |
| kevinbenton | ihrachys: let me make a note | 11:50 |
| kevinbenton | ihrachys: we'll probably get this in pecan side | 11:50 |
| openstackgerrit | John Schwarz proposed openstack/neutron: Mark an HA router as ALLOCATING before deletion https://review.openstack.org/348224 | 11:50 |
| openstackgerrit | Rabi Mishra proposed openstack/python-neutronclient: Add get_endpoint() method to v2 client https://review.openstack.org/348265 | 11:53 |
| *** tonytan4ever has joined #openstack-neutron | 11:53 | |
| *** lucasagomes is now known as lucas-hungry | 11:54 | |
| ihrachys | jlibosva: ajo: do you remember why from_db_object receives *objs? | 11:54 |
| ihrachys | that does not make sense | 11:54 |
| ihrachys | and I don't see any usage for multiple in code | 11:54 |
| *** chlong has joined #openstack-neutron | 11:55 | |
| ihrachys | unless you have some idea, I will just kill it | 11:55 |
| *** banix has quit IRC | 11:55 | |
| *** hoonetorg has quit IRC | 11:56 | |
| ihrachys | oh I think I remember! | 11:57 |
| *** tonytan4ever has quit IRC | 11:57 | |
| *** karthiks has joined #openstack-neutron | 11:58 | |
| ihrachys | I think it was at the times when we tried to implement qos rules with multiple models | 11:58 |
| ajo | ihrachys, I can't remember off my head, but I will look at it | 11:58 |
| ihrachys | one containing common attrs, and one specifc | 11:58 |
| openstackgerrit | Rabi Mishra proposed openstack/python-neutronclient: Add get_endpoint() method to v2 client https://review.openstack.org/348265 | 11:58 |
| ajo | I remember I reviewed the patch where it was introduced | 11:58 |
| * ajo reads | 11:58 | |
| ihrachys | ajo: do I make sense? | 11:58 |
| *** yuelongguang has joined #openstack-neutron | 11:58 | |
| ajo | ihrachys, I need to look at it | 11:58 |
| *** liuyulong_ has joined #openstack-neutron | 11:58 | |
| ajo | I can't remember | 11:58 |
| ajo | but, if it's not used, we can probably kill it | 11:59 |
| openstackgerrit | John Schwarz proposed openstack/neutron: Don't use exponential back-off for report_state https://review.openstack.org/347708 | 11:59 |
| *** iranzo has joined #openstack-neutron | 12:00 | |
| *** iranzo has joined #openstack-neutron | 12:00 | |
| *** lujinluo has quit IRC | 12:00 | |
| *** amuller has joined #openstack-neutron | 12:00 | |
| *** liuyulong has joined #openstack-neutron | 12:00 | |
| *** ramishra has joined #openstack-neutron | 12:02 | |
| jschwarz | liuyulong, reviewed https://review.openstack.org/#/c/265680/20 | 12:03 |
| jschwarz | liuyulong, I didn't run rally test - I will hopefully do it in the next hour or so | 12:03 |
| *** wolverineav has joined #openstack-neutron | 12:04 | |
| *** boden has joined #openstack-neutron | 12:04 | |
| *** liuyulong_ has quit IRC | 12:05 | |
| *** yuelongguang has quit IRC | 12:05 | |
| *** amotoki_ has quit IRC | 12:06 | |
| *** nyechiel has quit IRC | 12:07 | |
| openstackgerrit | Ihar Hrachyshka proposed openstack/neutron: objects: remove support for multiple db models in from_db_object https://review.openstack.org/348271 | 12:07 |
| *** mosulica has quit IRC | 12:08 | |
| *** wolverineav has quit IRC | 12:08 | |
| openstackgerrit | Takashi NATSUME proposed openstack/neutron: Fix a typo in neutron/services/trunk/rules.py https://review.openstack.org/346634 | 12:08 |
| *** kawa2014 has joined #openstack-neutron | 12:08 | |
| *** saggi has quit IRC | 12:11 | |
| *** trananhkma has quit IRC | 12:12 | |
| zioproto | is anyone using LBaaS ? I just realized I need to upgrade to LBaaS V2, but I have LBaaS V1 already deployed, there a migration path ? :) | 12:13 |
| *** amotoki has joined #openstack-neutron | 12:13 | |
| liuyulong | jschwarz, OK fine, thanks. | 12:13 |
| jlibosva | ihrachys: I wanted to ask on review, to comment why it's there. I don't know the reason. | 12:14 |
| jlibosva | ihrachys: we can try to remove it and we'll see what happens :) | 12:14 |
| ihrachys | jlibosva: I pushed the patch ^ | 12:17 |
| ihrachys | rossella_s: https://review.openstack.org/#/c/334380/15 | 12:18 |
| *** lixiaoy1 has quit IRC | 12:18 | |
| *** lixiaoy1 has joined #openstack-neutron | 12:19 | |
| *** jckasper has joined #openstack-neutron | 12:21 | |
| *** jhershbe has quit IRC | 12:24 | |
| *** edand has quit IRC | 12:24 | |
| *** jckasper has quit IRC | 12:26 | |
| openstackgerrit | Ihar Hrachyshka proposed openstack/neutron: objects: expose database model for NeutronDbObject instances https://review.openstack.org/348279 | 12:26 |
| ihrachys | kevinbenton: ^ exposing the model | 12:26 |
| ihrachys | note there is a dep | 12:26 |
| *** hoonetorg has joined #openstack-neutron | 12:27 | |
| ihrachys | kevinbenton: also if you have review cycles, would be great to unblock https://review.openstack.org/#/c/334380/ | 12:27 |
| ihrachys | it is in conflict with the model exposure patch, and I would better avoid introducing a conflict for korzen_'s piece of art | 12:27 |
| *** mkolesni has quit IRC | 12:29 | |
| *** wolverineav has joined #openstack-neutron | 12:30 | |
| *** nmagnezi_ has joined #openstack-neutron | 12:31 | |
| *** nmagnezi has quit IRC | 12:31 | |
| *** sindhu has quit IRC | 12:31 | |
| *** obondarev has joined #openstack-neutron | 12:32 | |
| *** yuelongguang has joined #openstack-neutron | 12:33 | |
| *** tpsilva has joined #openstack-neutron | 12:34 | |
| *** amotoki has quit IRC | 12:34 | |
| *** dave-mccowan has joined #openstack-neutron | 12:35 | |
| *** manikanta_tadi has quit IRC | 12:35 | |
| rossella_s | ihrachys, ack, thanks for the ping. merged! | 12:35 |
| ihrachys | woohoo | 12:36 |
| ihrachys | rossella_s: ^5! | 12:36 |
| *** salv-orlando has quit IRC | 12:36 | |
| *** wolverineav has quit IRC | 12:38 | |
| ajo | ihrachys, look at L63 here: https://review.openstack.org/#/c/347662/5/neutron/services/trunk/plugin.py@63 | 12:38 |
| ajo | interesting stuff, when did that happen? :) | 12:38 |
| *** amotoki has joined #openstack-neutron | 12:39 | |
| ajo | ihrachys, : | 12:39 |
| ajo | db_base_plugin_v2.NeutronDbPluginV2.register_dict_extend_funcs( | 12:39 |
| ajo | attributes.PORTS, [_extend_port_trunk_details]) | 12:39 |
| *** links has quit IRC | 12:40 | |
| *** yamamoto has quit IRC | 12:40 | |
| *** julim has joined #openstack-neutron | 12:43 | |
| *** saggi has joined #openstack-neutron | 12:46 | |
| *** amotoki has quit IRC | 12:47 | |
| liuyulong | jschwarz, ping | 12:52 |
| *** wolverineav has joined #openstack-neutron | 12:52 | |
| ihrachys | ajo: define 'that' | 12:52 |
| *** baoli has joined #openstack-neutron | 12:53 | |
| *** baoli_ has joined #openstack-neutron | 12:54 | |
| *** cleong has joined #openstack-neutron | 12:54 | |
| *** wolverineav has quit IRC | 12:57 | |
| *** yfried has quit IRC | 12:57 | |
| *** saggi has quit IRC | 12:57 | |
| *** pradk has quit IRC | 12:58 | |
| *** baoli has quit IRC | 12:58 | |
| *** nirmoy has quit IRC | 13:00 | |
| *** hoangcx has joined #openstack-neutron | 13:02 | |
| *** lucas-hungry is now known as lucasagomes | 13:03 | |
| jschwarz | liuyulong, hey | 13:05 |
| *** saggi has joined #openstack-neutron | 13:05 | |
| openstackgerrit | Ihar Hrachyshka proposed openstack/neutron: objects: expose database model for NeutronDbObject instances https://review.openstack.org/348279 | 13:06 |
| openstackgerrit | Ihar Hrachyshka proposed openstack/neutron: objects: remove support for multiple db models in from_db_object https://review.openstack.org/348271 | 13:06 |
| *** numans has quit IRC | 13:06 | |
| *** edmondsw has joined #openstack-neutron | 13:07 | |
| *** ramishra has quit IRC | 13:07 | |
| *** jckasper has joined #openstack-neutron | 13:07 | |
| *** roeyc has joined #openstack-neutron | 13:07 | |
| *** edmondsw has quit IRC | 13:08 | |
| jschwarz | obondarev, hey | 13:08 |
| jschwarz | obondarev, have a look at https://review.openstack.org/#/c/265672/10 - it tries to fix the l3 agent infinite loop issue | 13:09 |
| *** nmagnezi_ is now known as nmagnezi | 13:09 | |
| liuyulong | jschwarz, I've changed the https://review.openstack.org/#/c/265680 patch commit title, and add some comments to the code same as patch set 19. | 13:09 |
| jschwarz | liuyulong, maybe you have, but the server doesn't show any changes | 13:10 |
| jschwarz | liuyulong, did you push the changes? :) | 13:10 |
| openstackgerrit | LIU Yulong proposed openstack/neutron: Filter HA router without HA port bindings after race conditions https://review.openstack.org/265680 | 13:10 |
| obondarev | jschwarz: ah, thanks, will take a look | 13:10 |
| liuyulong | jschwarz, ... git review takes a long time ^.... | 13:10 |
| jschwarz | liuyulong, ok | 13:11 |
| jschwarz | liuyulong, please note that re: https://review.openstack.org/#/c/265672, obondarev has started also working on this exact issue so I suggest you coordinate with him | 13:11 |
| jschwarz | perhaps obondarev has detected more places that can be referenced without being initialized. | 13:11 |
| obondarev | jschwarz: I’m gonna fix a bit different issue | 13:12 |
| liuyulong | jschwarz, I've leave a comment to that https://bugs.launchpad.net/neutron/+bug/1606844. | 13:12 |
| openstack | Launchpad bug 1606844 in neutron "L3 agent constantly resyncing deleted router" [Medium,New] - Assigned to Oleg Bondarev (obondarev) | 13:12 |
| *** amotoki has joined #openstack-neutron | 13:12 | |
| liuyulong | jschwarz, https://bugs.launchpad.net/neutron/+bug/1533441 seems to be the right place to trace the issue. | 13:13 |
| openstack | Launchpad bug 1533441 in neutron "HA router can not be deleted in L3 agent after race between HA router creating and deleting" [Medium,Fix released] - Assigned to LIU Yulong (dragon889) | 13:13 |
| jschwarz | liuyulong, I would prefer not | 13:14 |
| *** zkassab has joined #openstack-neutron | 13:14 | |
| liuyulong | jschwarz, obondarev, is there now any patch on it ? | 13:14 |
| *** edmondsw has joined #openstack-neutron | 13:14 | |
| jschwarz | liuyulong, that bug report contains a few issues in its opening message and has multiple patches that claim to solve it | 13:14 |
| jschwarz | liuyulong, adding even more patches to it will make it very difficult to review your code | 13:14 |
| jschwarz | liuyulong, so lets stick to https://bugs.launchpad.net/neutron/+bug/1606844 | 13:15 |
| openstack | Launchpad bug 1606844 in neutron "L3 agent constantly resyncing deleted router" [Medium,New] - Assigned to Oleg Bondarev (obondarev) | 13:15 |
| jschwarz | ok? | 13:15 |
| liuyulong | jschwarz, i'm OK to have a new LP bug to solve it. | 13:15 |
| jschwarz | great | 13:15 |
| *** sindhu has joined #openstack-neutron | 13:15 | |
| *** gongysh has joined #openstack-neutron | 13:16 | |
| obondarev | jschwarz: lets keep 1606844 separate from HA router issues, I’m gonna upload a fix for it soon | 13:16 |
| *** oanson has quit IRC | 13:16 | |
| liuyulong | obondarev, If you have some time, maybe you can go through here https://review.openstack.org/#/c/265672, to see if you could help. | 13:16 |
| *** numans has joined #openstack-neutron | 13:17 | |
| jschwarz | obondarev, ack | 13:17 |
| *** nirmoy has joined #openstack-neutron | 13:17 | |
| obondarev | liuyulong: I quickly went through it and it doesn’t seem to fix the issue I’ve reported | 13:17 |
| *** akshai has joined #openstack-neutron | 13:18 | |
| obondarev | liuyulong: 1606844 is a pure l3 agent issue | 13:18 |
| jschwarz | liuyulong, so in continuation to what Oleg wrote, can you file a new bug report that contains reproduction steps and exact stacktraces of the issues you're trying to fix/ | 13:18 |
| *** emagana has joined #openstack-neutron | 13:19 | |
| *** bks has quit IRC | 13:19 | |
| *** wolverineav has joined #openstack-neutron | 13:19 | |
| liuyulong | obondarev, the HA router creating and deleting race condition is the main reason to that `infinitive loop` issue. Hope to see your patch. | 13:20 |
| liuyulong | jschwarz, OK, i will add a new LP bug. | 13:21 |
| obondarev | liuyulong: not only, I faced another case, it was non-HA router | 13:21 |
| jschwarz | liuyulong, thanks a lot :) | 13:22 |
| obondarev | liuyulong: generally it doesn’t make sense to retry router deletion if there is no router namespace | 13:22 |
| *** gouthamr has joined #openstack-neutron | 13:22 | |
| *** ramishra has joined #openstack-neutron | 13:23 | |
| liuyulong | obondarev, yeah, i've mentioned that at https://bugs.launchpad.net/neutron/+bug/1533441 "infinite loop 5 - 7" : ) | 13:24 |
| openstack | Launchpad bug 1533441 in neutron "HA router can not be deleted in L3 agent after race between HA router creating and deleting" [Medium,Fix released] - Assigned to LIU Yulong (dragon889) | 13:24 |
| *** emagana has quit IRC | 13:24 | |
| *** karthiks has quit IRC | 13:26 | |
| jlibosva | ihrachys: do you know if we have any xenial jobs that would contain information about the system - like e.g. kernel version? | 13:26 |
| *** banix has joined #openstack-neutron | 13:26 | |
| *** wolverineav has quit IRC | 13:26 | |
| ihrachys | jlibosva: not that I know. I think we worlddump in tempest/grenade only | 13:26 |
| ihrachys | afaik it's just docs/py* atm for xenial | 13:26 |
| jschwarz | liuyulong, so I tried to reproduce the original issue that https://review.openstack.org/#/c/265680/21 tries to fix | 13:27 |
| jschwarz | liuyulong, but I'm not successful | 13:27 |
| jschwarz | so i can't verify that this patch fixes it :\ | 13:27 |
| jlibosva | ihrachys: aha, so no experimental functional xenial or something? ok, thanks | 13:28 |
| *** ygbo has quit IRC | 13:28 | |
| *** jckasper has quit IRC | 13:28 | |
| liuyulong | jschwarz, we've talked about that in 265680, it's a really really small chance race. | 13:29 |
| *** jckasper has joined #openstack-neutron | 13:29 | |
| jlibosva | so I'll trust the release notes: Ubuntu 16.04 LTS is based on the long-term supported Linux release series 4.4. | 13:29 |
| jschwarz | liuyulong, yes. it actually reproduced several times earlier for me, but I restarted my setup and now it doesn't reproduce :< | 13:29 |
| *** ygbo has joined #openstack-neutron | 13:30 | |
| *** devvesa has quit IRC | 13:30 | |
| *** markvoelker has joined #openstack-neutron | 13:30 | |
| jschwarz | liuyulong, I'll try again with more concurrent threads... I'll be in touch soon | 13:30 |
| liuyulong | jschwarz, according to Kevin's DB opinion, if this line `if not port:` could work properly, then we can ensure that issue will be fix. | 13:31 |
| liuyulong | jschwarz, my rally config is always 40+ concurrency. | 13:33 |
| *** jckasper has quit IRC | 13:34 | |
| *** markvoelker has quit IRC | 13:34 | |
| *** ekuris has quit IRC | 13:35 | |
| *** gouthamr has quit IRC | 13:36 | |
| *** jamesdenton has joined #openstack-neutron | 13:36 | |
| *** aqkhan has quit IRC | 13:36 | |
| *** karthiks has joined #openstack-neutron | 13:38 | |
| *** yamamoto has joined #openstack-neutron | 13:41 | |
| liuyulong | jschwarz, a new bug was created here https://bugs.launchpad.net/neutron/+bug/1607381 | 13:41 |
| openstack | Launchpad bug 1607381 in neutron "HA router in l3 dvr_snat/legacy agent has no ha_port" [Undecided,New] | 13:41 |
| openstackgerrit | LIU Yulong proposed openstack/neutron: Clean up L3 agent side HA router stuffs https://review.openstack.org/265672 | 13:42 |
| *** ushkalim has quit IRC | 13:42 | |
| *** wolverineav has joined #openstack-neutron | 13:42 | |
| *** janzian has joined #openstack-neutron | 13:43 | |
| otherwiseguy | amuller: It doesn't look like the tests with the mocking are going to be stable. There's just too much stuff that changes behind the scenes. We can loop through thing above the post_commit function based on what's going on with ovsdb-server, etc. It's like going into a giant state machine and switching out the state and pretending you didn't (faking that vswitchd didn't compete, then looping anyway, etc.). | 13:44 |
| *** yamamoto has quit IRC | 13:46 | |
| *** itisha has joined #openstack-neutron | 13:46 | |
| *** crose has joined #openstack-neutron | 13:48 | |
| *** chandankumar has quit IRC | 13:50 | |
| *** wolverineav has quit IRC | 13:51 | |
| *** pgadiya has quit IRC | 13:51 | |
| *** gouthamr has joined #openstack-neutron | 13:51 | |
| *** tonytan4ever has joined #openstack-neutron | 13:52 | |
| *** saggi has quit IRC | 13:54 | |
| *** ramishra has quit IRC | 13:57 | |
| *** eilert has joined #openstack-neutron | 13:58 | |
| *** wolverineav has joined #openstack-neutron | 13:59 | |
| *** ddaskal has joined #openstack-neutron | 14:00 | |
| openstackgerrit | Merged openstack/neutron: objects: loading synthetic fields from defined ORM relationships. https://review.openstack.org/334380 | 14:00 |
| *** anilvenkata has quit IRC | 14:00 | |
| *** pradk has joined #openstack-neutron | 14:02 | |
| *** ramishra has joined #openstack-neutron | 14:02 | |
| otherwiseguy | amuller: I meant to write "aren't going to be stable" :p | 14:04 |
| *** bvandewa has joined #openstack-neutron | 14:05 | |
| *** ddaskal has quit IRC | 14:05 | |
| *** wolverineav has quit IRC | 14:05 | |
| *** a_ta has joined #openstack-neutron | 14:05 | |
| *** nlahouti has joined #openstack-neutron | 14:07 | |
| *** nlahouti1 has joined #openstack-neutron | 14:07 | |
| *** nlahouti has quit IRC | 14:07 | |
| *** mlavalle has joined #openstack-neutron | 14:08 | |
| openstackgerrit | Oleg Bondarev proposed openstack/neutron: L3 agent: check router namespace existence before delete https://review.openstack.org/348372 | 14:09 |
| *** edand has joined #openstack-neutron | 14:09 | |
| obondarev | jschwarz: liuyulong: ^^ | 14:09 |
| obondarev | still need to add tests | 14:09 |
| *** dane_leblanc has joined #openstack-neutron | 14:10 | |
| *** ajmiller has joined #openstack-neutron | 14:11 | |
| *** sindhu has quit IRC | 14:11 | |
| *** sridharg has quit IRC | 14:13 | |
| amuller | otherwiseguy: ack | 14:13 |
| openstackgerrit | Ihar Hrachyshka proposed openstack/neutron: tests: enable test_get_objects_queries_constant for trunk ports https://review.openstack.org/348378 | 14:14 |
| *** gongysh has quit IRC | 14:15 | |
| openstackgerrit | Oleg Bondarev proposed openstack/neutron: L3 agent: check router namespace existence before delete https://review.openstack.org/348372 | 14:15 |
| *** oshvartz has quit IRC | 14:16 | |
| *** sindhu has joined #openstack-neutron | 14:18 | |
| jschwarz | obondarev, ack | 14:18 |
| jschwarz | liuyulong, ping | 14:20 |
| liuyulong | jschwarz, pong | 14:20 |
| liuyulong | obondarev, hi, i'm looking at that patch. | 14:21 |
| jschwarz | liuyulong, so the second stacktrace you posted, http://paste.openstack.org/show/528407/, is relevant to what obondarev is trying to fix | 14:21 |
| jschwarz | liuyulong, it's not something that is related to your bug at all | 14:21 |
| liuyulong | jschwarz, ha_port absent could raise that trace. | 14:22 |
| openstackgerrit | Merged openstack/neutron: Skip DHCP provisioning block for network ports https://review.openstack.org/346323 | 14:22 |
| jschwarz | liuyulong, it could, but that's a side effect of http://paste.openstack.org/show/523757/ | 14:23 |
| jschwarz | liuyulong, correct me if I'm wrong: the first iteration of the infinite loop goes on, deletes the namespace and then fails with that trace (http://paste.openstack.org/show/523757/) | 14:24 |
| liuyulong | jschwarz, OK, I move that to 1606844. | 14:24 |
| jschwarz | liuyulong, then, the second iteration will fail with "no namespace" (which is what Oleg is fixing) | 14:24 |
| jschwarz | liuyulong, ok :) | 14:24 |
| liuyulong | jschwarz, yeah | 14:24 |
| jschwarz | liuyulong, I just want to make everything as clear as possible because otherwise people work on the same things and the bug reports are very messy | 14:25 |
| jschwarz | :) | 14:25 |
| liuyulong | jschwarz, lol | 14:25 |
| jschwarz | liuyulong, it does | 14:26 |
| *** wolverineav has joined #openstack-neutron | 14:26 | |
| jschwarz | liuyulong, it took me 4 hours yesterday to go over some 20 different bugs that were talking about like 15 issues | 14:27 |
| jschwarz | and 5 were simply duplicates of others and weren't marked as such | 14:27 |
| *** mattgreene has joined #openstack-neutron | 14:27 | |
| liuyulong | obondarev, please make sure that if skip namespace delete, that the router_info deletion could go further properly. And I'm OK with this ns.exists() check now. + 1 | 14:27 |
| *** boden has quit IRC | 14:27 | |
| *** saggi has joined #openstack-neutron | 14:27 | |
| liuyulong | jschwarz, so much details in neutron, : ) | 14:28 |
| jschwarz | liuyulong, encountered your issue with 40 concurrent threads | 14:28 |
| *** Alex_Stef has quit IRC | 14:28 | |
| jschwarz | liuyulong, will now apply your fix and try to see if it happens again | 14:28 |
| liuyulong | jschwarz, great | 14:28 |
| *** itzikb has quit IRC | 14:29 | |
| *** wolverineav has quit IRC | 14:31 | |
| *** wolverineav has joined #openstack-neutron | 14:32 | |
| liuyulong | jschwarz, so now, how many LP bug about HA router race condition do we have now ? | 14:33 |
| *** nyechiel has joined #openstack-neutron | 14:35 | |
| *** jistr is now known as jistr|call | 14:35 | |
| jschwarz | liuyulong, I need to do a re-count after today | 14:35 |
| jschwarz | liuyulong, probably around 9? | 14:35 |
| jschwarz | liuyulong, also, my setup crashed a bit so I need to re-start it | 14:36 |
| *** matrohon has quit IRC | 14:36 | |
| *** ramishra has quit IRC | 14:36 | |
| *** nlahouti1 has quit IRC | 14:36 | |
| *** nlahouti has joined #openstack-neutron | 14:37 | |
| *** ramishra has joined #openstack-neutron | 14:37 | |
| *** janzian has quit IRC | 14:37 | |
| ajo | ihrachys, https://review.openstack.org/#/c/347662/5/neutron/services/trunk/rpc/server.py@L66 if you can. Have we ever used osloversioned objects remotable methods yet? | 14:37 |
| *** nlahouti has quit IRC | 14:37 | |
| ihrachys | ajo: I have not looked into those | 14:38 |
| *** pbandark has quit IRC | 14:38 | |
| ihrachys | ajo: what's the use case for an agent to update trunk? | 14:38 |
| *** wolverineav has quit IRC | 14:38 | |
| ajo | ihrachys, seems interesting, could solve some use cases I'm seeing in those patches | 14:38 |
| ajo | ihrachys, I don't know, may be to provide status of some kind? | 14:39 |
| *** marst has quit IRC | 14:39 | |
| *** nmagnezi has quit IRC | 14:39 | |
| *** vijaykc4 has joined #openstack-neutron | 14:40 | |
| liuyulong | jschwarz, now i have 4, https://bugs.launchpad.net/neutron/+bug/1533443, https://bugs.launchpad.net/neutron/+bug/1533457, https://bugs.launchpad.net/neutron/+bug/1533455, https://bugs.launchpad.net/neutron/+bug/1607381 | 14:40 |
| openstack | Launchpad bug 1533443 in neutron "ML2: cannot update HA router ha_port states after race between HA router creating and deleting" [Low,In progress] - Assigned to LIU Yulong (dragon889) | 14:40 |
| openstack | Launchpad bug 1533457 in neutron "Neutron server unable to sync HA info after race between HA router creating and deleting" [Medium,In progress] - Assigned to LIU Yulong (dragon889) | 14:40 |
| openstack | Launchpad bug 1533455 in neutron "Stale processes lives after a fanout deleting HA router RPC between L3 agents" [Medium,In progress] - Assigned to LIU Yulong (dragon889) | 14:40 |
| openstack | Launchpad bug 1607381 in neutron "HA router in l3 dvr_snat/legacy agent has no ha_port" [Undecided,In progress] - Assigned to LIU Yulong (dragon889) | 14:40 |
| jschwarz | liuyulong, those are only the ones that are assigned to you | 14:41 |
| jschwarz | liuyulong, some are also assigned to me, to Oleg and to Ann | 14:41 |
| jschwarz | I have 2, Ann has 3 and Oleg as few | 14:41 |
| liuyulong | jschwarz, OK, please send the email to me about that. | 14:41 |
| *** clenimar has joined #openstack-neutron | 14:42 | |
| liuyulong | jschwarz, It's almost 11:00 p.m. in Beijing, I need to take a bath. See you then. | 14:43 |
| *** liuyulong has quit IRC | 14:44 | |
| openstackgerrit | Ihar Hrachyshka proposed openstack/neutron: tests: enable test_get_objects_queries_constant for trunk ports https://review.openstack.org/348378 | 14:44 |
| openstackgerrit | Ihar Hrachyshka proposed openstack/neutron: trunk: avoid redundant refetch of subports on create https://review.openstack.org/348396 | 14:44 |
| openstackgerrit | Ihar Hrachyshka proposed openstack/neutron: trunk: declare port_id as a primary key https://review.openstack.org/348397 | 14:44 |
| openstackgerrit | Ihar Hrachyshka proposed openstack/neutron: tests: check that trunk sub_ports field is properly populated https://review.openstack.org/348398 | 14:44 |
| *** dasanind has joined #openstack-neutron | 14:45 | |
| *** vijaykc4 has quit IRC | 14:45 | |
| *** janzian has joined #openstack-neutron | 14:46 | |
| *** wolverineav has joined #openstack-neutron | 14:46 | |
| *** ijw has joined #openstack-neutron | 14:47 | |
| *** ijw has quit IRC | 14:47 | |
| *** ijw has joined #openstack-neutron | 14:47 | |
| *** vijaykc4 has joined #openstack-neutron | 14:48 | |
| *** jistr|call is now known as jistr | 14:49 | |
| openstackgerrit | Pablo Iranzo Gómez proposed openstack/neutron-lib: Enhance valid_values validations to check that valid_values has method __contains__ https://review.openstack.org/343799 | 14:49 |
| *** wolverineav has quit IRC | 14:50 | |
| *** vijaykc4 has quit IRC | 14:51 | |
| *** ijw has quit IRC | 14:52 | |
| *** wolverineav has joined #openstack-neutron | 14:52 | |
| *** nmagnezi has joined #openstack-neutron | 14:52 | |
| *** nlahouti has joined #openstack-neutron | 14:53 | |
| *** vhoward has joined #openstack-neutron | 14:54 | |
| *** nlahouti has quit IRC | 14:54 | |
| *** Swami has joined #openstack-neutron | 14:55 | |
| *** Swami_ has joined #openstack-neutron | 14:55 | |
| *** marst has joined #openstack-neutron | 14:55 | |
| *** nlahouti has joined #openstack-neutron | 14:55 | |
| *** tyrola has quit IRC | 14:55 | |
| *** vikram has quit IRC | 14:56 | |
| *** emagana has joined #openstack-neutron | 14:57 | |
| *** EinstCra_ has joined #openstack-neutron | 14:57 | |
| *** garyk has quit IRC | 14:58 | |
| *** garyk has joined #openstack-neutron | 14:58 | |
| *** saggi has quit IRC | 14:59 | |
| *** numans has quit IRC | 14:59 | |
| *** nlahouti has quit IRC | 15:00 | |
| *** Swami__ has joined #openstack-neutron | 15:01 | |
| openstackgerrit | Ryan Tidwell proposed openstack/neutron: Enable adoption of subnets into a subnet pool https://review.openstack.org/348080 | 15:03 |
| *** iyamahat has joined #openstack-neutron | 15:03 | |
| *** johnbelamaric has joined #openstack-neutron | 15:04 | |
| *** zioproto has quit IRC | 15:04 | |
| *** Swami has quit IRC | 15:05 | |
| *** EinstCra_ has quit IRC | 15:06 | |
| *** yamamoto has joined #openstack-neutron | 15:06 | |
| *** EinstCrazy has joined #openstack-neutron | 15:06 | |
| *** hynekm has quit IRC | 15:07 | |
| nmagnezi | jlibosva, https://review.openstack.org/#/c/345263/11/neutron/tests/unit/agent/linux/test_external_process.py | 15:08 |
| *** Swami has joined #openstack-neutron | 15:08 | |
| nmagnezi | jlibosva, not sure i'm following here.. | 15:08 |
| jlibosva | nmagnezi: you're passing a namespace to ExternalProcess while you could omit that and use root namespace | 15:08 |
| jlibosva | nmagnezi: this would allow you to use just self.execute without mocking IpNetnsCommand | 15:09 |
| *** armax has joined #openstack-neutron | 15:09 | |
| jlibosva | nmagnezi: as the fact that if you use namespace then command is ran in namespace isn't subject to test in the test you wrote | 15:10 |
| *** yamamoto has quit IRC | 15:10 | |
| *** emagana has quit IRC | 15:10 | |
| jlibosva | nmagnezi: makes sense now? | 15:10 |
| *** Swami__ has quit IRC | 15:11 | |
| *** ddaskal has joined #openstack-neutron | 15:12 | |
| *** EinstCrazy has quit IRC | 15:12 | |
| HenryG | dasm: good morning | 15:16 |
| *** wolverineav has quit IRC | 15:16 | |
| HenryG | dasm: Please rebase https://review.openstack.org/344383 | 15:17 |
| *** emagana has joined #openstack-neutron | 15:17 | |
| *** wolverineav has joined #openstack-neutron | 15:18 | |
| *** openstackgerrit has quit IRC | 15:18 | |
| *** ijw has joined #openstack-neutron | 15:18 | |
| *** openstackgerrit has joined #openstack-neutron | 15:19 | |
| *** yuanying_ has joined #openstack-neutron | 15:20 | |
| *** pcaruana has quit IRC | 15:22 | |
| *** yuanying has quit IRC | 15:22 | |
| *** ddaskal has quit IRC | 15:22 | |
| *** ijw has quit IRC | 15:23 | |
| *** wolverineav has quit IRC | 15:24 | |
| *** kbringard has joined #openstack-neutron | 15:26 | |
| *** bvandewa has quit IRC | 15:27 | |
| *** wolverineav has joined #openstack-neutron | 15:27 | |
| *** jckasper has joined #openstack-neutron | 15:30 | |
| *** edand has quit IRC | 15:30 | |
| *** elo has quit IRC | 15:30 | |
| nmagnezi | jlibosva, yes. now I get it. thanks a lot Jakub | 15:31 |
| jlibosva | nmagnezi: np | 15:31 |
| *** slunkad_ has joined #openstack-neutron | 15:32 | |
| *** elo has joined #openstack-neutron | 15:32 | |
| *** slunkad_ has quit IRC | 15:34 | |
| *** emagana has quit IRC | 15:34 | |
| *** ijw has joined #openstack-neutron | 15:34 | |
| *** emagana has joined #openstack-neutron | 15:35 | |
| openstackgerrit | Terry Wilson proposed openstack/neutron: Wait for vswitchd to add interfaces in native ovsdb https://review.openstack.org/344859 | 15:36 |
| otherwiseguy | amuller: kevinbenton: ^ I think I've addressed concerns and removed tests that relied on flaky timeout_exceeded side_effects. | 15:37 |
| *** Swami__ has joined #openstack-neutron | 15:37 | |
| *** charliekang has joined #openstack-neutron | 15:37 | |
| *** mhickey has quit IRC | 15:38 | |
| *** Swami__ has quit IRC | 15:38 | |
| *** Swami__ has joined #openstack-neutron | 15:38 | |
| *** iranzo has quit IRC | 15:38 | |
| *** ijw has quit IRC | 15:39 | |
| *** bvandewa has joined #openstack-neutron | 15:39 | |
| *** bvandewa has quit IRC | 15:39 | |
| *** mickeys has joined #openstack-neutron | 15:39 | |
| *** bvandewa has joined #openstack-neutron | 15:39 | |
| *** Swami has quit IRC | 15:40 | |
| *** Swami_ has quit IRC | 15:40 | |
| amuller | otherwiseguy: did you run the entire test suite locally a few times? | 15:42 |
| *** jistr is now known as jistr|afk | 15:42 | |
| amuller | otherwiseguy: I don't want to trade one form of gate instability for another =p | 15:42 |
| *** zhhuabj has quit IRC | 15:42 | |
| otherwiseguy | I ran the ovs_lib ones several times. I ran the new ones several thousand times. | 15:42 |
| jlibosva | nmagnezi: comments, comments, comments :) | 15:42 |
| openstackgerrit | Armando Migliaccio proposed openstack/neutron: Add RPC layer for Trunk Plugin and initial Open vSwitch driver https://review.openstack.org/347662 | 15:43 |
| otherwiseguy | amuller: the latest run of test_impl_idl test runs is currently on #2749. | 15:44 |
| amuller | otherwiseguy: ok let's wait for 2750 to be really sure. You know what they say... | 15:44 |
| amuller | There's nothing like the 2750th time | 15:44 |
| otherwiseguy | #2756 and counting | 15:44 |
| otherwiseguy | :) | 15:44 |
| *** jckasper has quit IRC | 15:46 | |
| *** iranzo has joined #openstack-neutron | 15:47 | |
| *** jckasper has joined #openstack-neutron | 15:47 | |
| openstackgerrit | John Schwarz proposed openstack/neutron: Fix a race with auto_schedule of HA routers https://review.openstack.org/284400 | 15:48 |
| *** jckasper has quit IRC | 15:48 | |
| *** itzikb has joined #openstack-neutron | 15:49 | |
| *** jckasper has joined #openstack-neutron | 15:49 | |
| *** zhhuabj has joined #openstack-neutron | 15:49 | |
| *** ijw has joined #openstack-neutron | 15:50 | |
| *** kobis has quit IRC | 15:51 | |
| *** wolverineav has quit IRC | 15:51 | |
| *** emagana has quit IRC | 15:51 | |
| *** emagana has joined #openstack-neutron | 15:52 | |
| *** wolverineav has joined #openstack-neutron | 15:52 | |
| *** comstud has quit IRC | 15:55 | |
| *** iranzo has quit IRC | 15:55 | |
| *** ijw has quit IRC | 15:55 | |
| *** nlahouti has joined #openstack-neutron | 15:56 | |
| *** anilvenkata has joined #openstack-neutron | 15:57 | |
| *** nlahouti1 has joined #openstack-neutron | 15:58 | |
| *** nlahouti has quit IRC | 15:58 | |
| *** pece has quit IRC | 15:58 | |
| openstackgerrit | Nir Magnezi proposed openstack/neutron: Adds a default reload callback to ProcessManager https://review.openstack.org/345263 | 15:59 |
| nmagnezi | jlibosva, replies, replies, replies :) | 15:59 |
| *** rossella_s has quit IRC | 16:00 | |
| *** rossella_s has joined #openstack-neutron | 16:00 | |
| *** Swami__ has quit IRC | 16:00 | |
| *** Leom has joined #openstack-neutron | 16:00 | |
| *** Leom has quit IRC | 16:01 | |
| haleyb | john-davidge: so do you see the cli still being --service-type foo ? i guess that's where i'm confused | 16:02 |
| openstackgerrit | Nir Magnezi proposed openstack/neutron: Adds a default reload callback to ProcessManager https://review.openstack.org/345263 | 16:03 |
| john-davidge | haleyb: Yeah, that shouldn't need to change since the user just uses that option multiple times to add more than one | 16:03 |
| john-davidge | but the property it passes to subnet create/update will need to be service_subnets rather than service_subnet | 16:03 |
| haleyb | service_types, got it | 16:04 |
| *** boden has joined #openstack-neutron | 16:05 | |
| john-davidge | haleyb: Haha, yep. I can't brain today. | 16:06 |
| *** david-lyle has joined #openstack-neutron | 16:07 | |
| *** diga has joined #openstack-neutron | 16:08 | |
| *** johnbelamaric has left #openstack-neutron | 16:09 | |
| *** Leom has joined #openstack-neutron | 16:09 | |
| *** Leom has quit IRC | 16:10 | |
| *** emagana has quit IRC | 16:11 | |
| *** Leom has joined #openstack-neutron | 16:12 | |
| *** obondarev has quit IRC | 16:13 | |
| *** kawa2014 has quit IRC | 16:13 | |
| *** emagana has joined #openstack-neutron | 16:14 | |
| *** Leom has quit IRC | 16:14 | |
| *** jhershbe has joined #openstack-neutron | 16:15 | |
| *** wolverineav has quit IRC | 16:16 | |
| *** Leom has joined #openstack-neutron | 16:17 | |
| *** zhhuabj has quit IRC | 16:19 | |
| *** baoli_ has quit IRC | 16:19 | |
| *** Leom_ has joined #openstack-neutron | 16:20 | |
| *** chandankumar has joined #openstack-neutron | 16:20 | |
| *** ijw has joined #openstack-neutron | 16:21 | |
| *** obondarev has joined #openstack-neutron | 16:21 | |
| *** wolverineav has joined #openstack-neutron | 16:22 | |
| openstackgerrit | Sindhu Devale proposed openstack/neutron: Refactoring agent linux&ovsdb config https://review.openstack.org/347867 | 16:22 |
| *** Leom has quit IRC | 16:23 | |
| *** ygbo has quit IRC | 16:24 | |
| *** jlanoux has quit IRC | 16:24 | |
| *** roeyc has quit IRC | 16:24 | |
| *** ijw has quit IRC | 16:26 | |
| openstackgerrit | John Schwarz proposed openstack/neutron: Fix a race with auto_schedule of HA routers https://review.openstack.org/284400 | 16:26 |
| *** mattgreene has quit IRC | 16:27 | |
| *** sbalukoff has quit IRC | 16:28 | |
| *** moshele has quit IRC | 16:29 | |
| *** emagana has quit IRC | 16:29 | |
| *** emagana has joined #openstack-neutron | 16:29 | |
| *** jistr|afk is now known as jistr | 16:30 | |
| *** kobis has joined #openstack-neutron | 16:31 | |
| *** tesseract- has quit IRC | 16:32 | |
| *** wolverineav has quit IRC | 16:34 | |
| *** wolverineav has joined #openstack-neutron | 16:35 | |
| *** kbringard has quit IRC | 16:37 | |
| *** kbringard has joined #openstack-neutron | 16:38 | |
| *** iyamahat has quit IRC | 16:41 | |
| *** wolverineav has quit IRC | 16:41 | |
| *** vhoward has quit IRC | 16:42 | |
| *** jhershbe has quit IRC | 16:42 | |
| openstackgerrit | Armando Migliaccio proposed openstack/neutron: Remove local subports validator https://review.openstack.org/348472 | 16:42 |
| *** wolverineav has joined #openstack-neutron | 16:46 | |
| *** kbringard has quit IRC | 16:46 | |
| *** hynekm has joined #openstack-neutron | 16:47 | |
| rtheis | armax: when you have time, http://eavesdrop.openstack.org/meetings/openstackclient/2016/openstackclient.2016-07-28-13.01.log.html provides some answers to your questions from the other day on osc plugins. | 16:48 |
| armax | rtheis: yes, it’s on my list | 16:48 |
| armax | rtheis: I saw your comments on the client patch fo trunks, much appreciate it | 16:49 |
| rtheis | yw | 16:49 |
| *** dasanind has quit IRC | 16:49 | |
| *** dasanind has joined #openstack-neutron | 16:49 | |
| armax | rtheis: so I guess the course of action is, ‘add todos, revisit later’? | 16:49 |
| *** emagana has quit IRC | 16:49 | |
| armax | rtheis: is that the gist of it? :) | 16:49 |
| *** emagana_ has joined #openstack-neutron | 16:49 | |
| armax | rtheis: I only glanced it over, but I was going to go in more details | 16:49 |
| *** Swami has joined #openstack-neutron | 16:50 | |
| rtheis | armax: yes, duplicate code not yet in osc-lib with todos to pull out once moved over to osc-lib | 16:50 |
| armax | rtheis: ack | 16:50 |
| *** imcsk8|zZz is now known as imcsk8 | 16:51 | |
| mfranc213_ | ping ajo | 16:51 |
| *** numans has joined #openstack-neutron | 16:54 | |
| *** banix has quit IRC | 16:54 | |
| *** amotoki has quit IRC | 16:54 | |
| mfranc213_ | ping ihar, davidsha | 16:55 |
| *** wolverineav has quit IRC | 16:55 | |
| *** sambetts is now known as sambetts|afk | 16:56 | |
| *** kbringard has joined #openstack-neutron | 16:57 | |
| *** wolverineav has joined #openstack-neutron | 16:59 | |
| *** iyamahat has joined #openstack-neutron | 16:59 | |
| mfranc213_ | hello ajo ihar davidsha: i wonder if you would be able to comment on https://review.openstack.org/#/c/339246/10/neutron/agent/l3/agent.py@433? | 16:59 |
| *** amotoki has joined #openstack-neutron | 17:00 | |
| *** sbalukoff has joined #openstack-neutron | 17:00 | |
| *** yamahata has joined #openstack-neutron | 17:00 | |
| *** jlibosva has quit IRC | 17:02 | |
| openstackgerrit | Miguel Angel Ajo proposed openstack/neutron: Introduce bulk_push to rpc callback mechanism https://review.openstack.org/348476 | 17:03 |
| openstackgerrit | Ihar Hrachyshka proposed openstack/neutron: objects; avoid additional fetch for prefixes on pool get https://review.openstack.org/348478 | 17:03 |
| *** david-lyle has quit IRC | 17:03 | |
| openstackgerrit | Ihar Hrachyshka proposed openstack/neutron: objects: avoid additional fetch for prefixes on pool get https://review.openstack.org/348478 | 17:03 |
| *** amotoki has quit IRC | 17:03 | |
| *** wolverineav has quit IRC | 17:03 | |
| ajo | mfranc213_, I will review tomorrow | 17:04 |
| ajo | I have to run now | 17:04 |
| ajo | I just read your discussion and I need to think about it. | 17:05 |
| *** abregman has quit IRC | 17:05 | |
| mfranc213_ | thank you ajo | 17:05 |
| ajo | I'd listen to carl, we could have differntiated calls | 17:05 |
| ajo | if there's a good reason for it | 17:06 |
| *** nmagnezi has quit IRC | 17:06 | |
| ajo | or even non-differentiated and differentiated | 17:06 |
| ajo | which the l2-agent extensions could eventually evolve onto | 17:06 |
| *** emagana_ has quit IRC | 17:06 | |
| *** emagana has joined #openstack-neutron | 17:06 | |
| mfranc213_ | it made sense to me. i was looking through qos, fwaas, and fdb code to see if i could se a reason for it | 17:06 |
| mfranc213_ | but a quick scan didn't surface anything for me | 17:07 |
| mfranc213_ | s/se/see | 17:07 |
| *** gvrangan has joined #openstack-neutron | 17:07 | |
| *** elopez has joined #openstack-neutron | 17:07 | |
| manjeets_ | ihrachys, ping | 17:08 |
| ihrachys | manjeets_: pong | 17:08 |
| *** elopez is now known as Guest10226 | 17:08 | |
| *** wolverineav has joined #openstack-neutron | 17:08 | |
| *** Guest10226 has quit IRC | 17:08 | |
| manjeets_ | i am starting refactor of models I have suggested 3 approaches | 17:08 |
| *** elopez_ has joined #openstack-neutron | 17:08 | |
| manjeets_ | ihrachys, I've sent email last night I feel second one is good | 17:09 |
| garyk | armax: can you please give this your blessing - https://review.openstack.org/348123 | 17:10 |
| *** akshai has quit IRC | 17:10 | |
| *** anilvenkata has quit IRC | 17:11 | |
| armax | garyk: ack | 17:11 |
| *** kevo has joined #openstack-neutron | 17:11 | |
| *** mattgreene has joined #openstack-neutron | 17:11 | |
| manjeets_ | ihrachys, just read your response for some reason my filters missed it | 17:12 |
| *** akshai has joined #openstack-neutron | 17:12 | |
| *** amuller is now known as amuller_afk | 17:12 | |
| garyk | armax: gracias! | 17:12 |
| *** cdl_ has joined #openstack-neutron | 17:14 | |
| *** banix has joined #openstack-neutron | 17:15 | |
| *** jckasper has quit IRC | 17:17 | |
| *** jckasper has joined #openstack-neutron | 17:17 | |
| *** fragatina has joined #openstack-neutron | 17:17 | |
| openstackgerrit | Ihar Hrachyshka proposed openstack/neutron: tests: check that trunk sub_ports field is properly populated https://review.openstack.org/348398 | 17:18 |
| *** sputnik13 has joined #openstack-neutron | 17:18 | |
| *** fragatina has quit IRC | 17:19 | |
| *** nherciu has quit IRC | 17:19 | |
| *** wolverineav has quit IRC | 17:20 | |
| *** s3wong has joined #openstack-neutron | 17:20 | |
| *** wolverineav has joined #openstack-neutron | 17:20 | |
| *** baoli has joined #openstack-neutron | 17:22 | |
| *** Leom_ has quit IRC | 17:22 | |
| *** emagana has quit IRC | 17:22 | |
| *** emagana has joined #openstack-neutron | 17:22 | |
| openstackgerrit | Carl Baldwin proposed openstack/neutron: Switch to pluggable IPAM implementation https://review.openstack.org/181023 | 17:24 |
| openstackgerrit | Carl Baldwin proposed openstack/neutron: Fix updating allocation_pools on subnet update https://review.openstack.org/345498 | 17:24 |
| *** fragatina has joined #openstack-neutron | 17:26 | |
| *** fragatina has quit IRC | 17:27 | |
| *** cdl_ has quit IRC | 17:27 | |
| *** mattgree_ has joined #openstack-neutron | 17:27 | |
| *** abhiraut has joined #openstack-neutron | 17:28 | |
| *** david-lyle has joined #openstack-neutron | 17:28 | |
| openstackgerrit | Carl Baldwin proposed openstack/neutron: Fix updating allocation_pools on subnet update https://review.openstack.org/345498 | 17:29 |
| *** emagana has quit IRC | 17:30 | |
| *** emagana has joined #openstack-neutron | 17:30 | |
| openstackgerrit | Carl Baldwin proposed openstack/neutron: Switch to pluggable IPAM implementation https://review.openstack.org/181023 | 17:31 |
| *** mattgreene has quit IRC | 17:31 | |
| *** john-davidge has quit IRC | 17:35 | |
| *** Swami has quit IRC | 17:36 | |
| *** Swami has joined #openstack-neutron | 17:37 | |
| *** akshai has quit IRC | 17:39 | |
| *** fragatina has joined #openstack-neutron | 17:39 | |
| *** fragatina has quit IRC | 17:40 | |
| *** fragatina has joined #openstack-neutron | 17:41 | |
| openstackgerrit | Ihar Hrachyshka proposed openstack/neutron: objects: expose database model for NeutronDbObject instances https://review.openstack.org/348279 | 17:44 |
| *** andymaier has quit IRC | 17:45 | |
| *** kbringard has quit IRC | 17:46 | |
| *** fragatina has quit IRC | 17:46 | |
| *** akshai has joined #openstack-neutron | 17:47 | |
| *** tidwellr has joined #openstack-neutron | 17:48 | |
| *** banix has quit IRC | 17:49 | |
| *** ihrachys has quit IRC | 17:50 | |
| *** hynekm has quit IRC | 17:50 | |
| *** emagana has quit IRC | 17:50 | |
| *** emagana has joined #openstack-neutron | 17:50 | |
| *** emagana has quit IRC | 17:55 | |
| *** chandankumar has quit IRC | 17:56 | |
| openstackgerrit | Manjeet Singh Bhatia proposed openstack/neutron: Add OVO for dns Objects https://review.openstack.org/334695 | 17:58 |
| *** vijaykc4 has joined #openstack-neutron | 17:59 | |
| *** akshai has quit IRC | 18:00 | |
| *** vijaykc4 has quit IRC | 18:00 | |
| *** akshai has joined #openstack-neutron | 18:01 | |
| *** mattgree_ has quit IRC | 18:01 | |
| *** emagana has joined #openstack-neutron | 18:02 | |
| *** mattgreene has joined #openstack-neutron | 18:02 | |
| *** emagana has quit IRC | 18:03 | |
| *** wolverineav has quit IRC | 18:04 | |
| *** nyechiel has quit IRC | 18:04 | |
| *** wolverineav has joined #openstack-neutron | 18:04 | |
| *** dasanind has quit IRC | 18:05 | |
| dasm | HenryG: thanks for information. Patch rebased thanks to electrocucaracha. | 18:07 |
| *** Sukhdev has joined #openstack-neutron | 18:07 | |
| *** yuelongguang_ has joined #openstack-neutron | 18:08 | |
| *** mohankumar has quit IRC | 18:08 | |
| HenryG | dasm: electrocucaracha: thanks, reviewing | 18:08 |
| electrocucaracha | HenryG: dasm I tried to find a middle point in that change hopefully that was the solution | 18:09 |
| *** amuller_afk is now known as amuller | 18:09 | |
| dasm | electrocucaracha: you have leftovers after merge :) i'm on it | 18:09 |
| *** yuelongguang has quit IRC | 18:09 | |
| *** yuelongguang_ is now known as yuelongguang | 18:10 | |
| *** Sukhdev has quit IRC | 18:10 | |
| openstackgerrit | Merged openstack/neutron-specs: LBaaS project spinout https://review.openstack.org/310805 | 18:10 |
| HenryG | dasm: electrocucaracha: commented | 18:11 |
| *** wolverineav has quit IRC | 18:12 | |
| electrocucaracha | :S | 18:12 |
| electrocucaracha | I forgot to remove some merge lines | 18:12 |
| *** obondarev has quit IRC | 18:13 | |
| *** ociuhandu has quit IRC | 18:14 | |
| dasm | HenryG: done. electrocucaracha, thanks for doing this :) | 18:14 |
| *** ajmiller has quit IRC | 18:14 | |
| electrocucaracha | dasm: wfh? | 18:15 |
| dasm | electrocucaracha: kind of. i was on volunteer meeting (SAPA). just got back from it. | 18:16 |
| *** yuelongguang has quit IRC | 18:16 | |
| *** Sukhdev has joined #openstack-neutron | 18:17 | |
| *** nlahouti1 has quit IRC | 18:19 | |
| *** nlahouti has joined #openstack-neutron | 18:19 | |
| *** rubasov has quit IRC | 18:20 | |
| *** yuelongguang has joined #openstack-neutron | 18:22 | |
| *** 7YUABPPH1 has joined #openstack-neutron | 18:22 | |
| *** 14WAACCVI has joined #openstack-neutron | 18:22 | |
| openstackgerrit | Rawlin Peters proposed openstack/neutron: Pass bridge_name in OVS port's vif_details https://review.openstack.org/343816 | 18:23 |
| *** claudiub|2 has joined #openstack-neutron | 18:23 | |
| dasm | HenryG: how did you know which one needs to be nullable and which one not? This is established by depends-on? | 18:24 |
| *** banix has joined #openstack-neutron | 18:25 | |
| *** 14WAACCVI has quit IRC | 18:25 | |
| *** 7YUABPPH1 has quit IRC | 18:25 | |
| *** obondarev has joined #openstack-neutron | 18:26 | |
| *** yfried has joined #openstack-neutron | 18:27 | |
| HenryG | dasm: no, it must match the schema created by the migrations | 18:28 |
| HenryG | dasm: https://github.com/openstack/networking-bgpvpn/blob/master/networking_bgpvpn/neutron/db/migration/alembic_migrations/versions/liberty/expand/17d9fd4fddee_initial.py#L40 | 18:28 |
| dasm | HenryG: True, you're right. I stopped thinking after adding this NotNullable class. | 18:28 |
| HenryG | dasm: there is a problem with the depends-on though, trying to figure it out ... | 18:29 |
| *** baoli has quit IRC | 18:29 | |
| dasm | HenryG: are you talking about neutron patch with bunch of depends-on? | 18:29 |
| HenryG | dasm: no, the alembic depends-on in your bgpvpn patch | 18:30 |
| *** obondarev has quit IRC | 18:30 | |
| HenryG | dasm: the test_migrations doesn't like it | 18:30 |
| dasm | ah. | 18:30 |
| *** dasanind has joined #openstack-neutron | 18:30 | |
| *** vijaykc4 has joined #openstack-neutron | 18:31 | |
| *** zhhuabj has joined #openstack-neutron | 18:32 | |
| *** vishwanathj has joined #openstack-neutron | 18:34 | |
| *** catintheroof has joined #openstack-neutron | 18:34 | |
| *** nlahouti has quit IRC | 18:37 | |
| *** nlahouti has joined #openstack-neutron | 18:37 | |
| *** tonytan4ever has quit IRC | 18:38 | |
| *** gvrangan has quit IRC | 18:38 | |
| *** Leom has joined #openstack-neutron | 18:40 | |
| *** nlahouti1 has joined #openstack-neutron | 18:41 | |
| *** nlahouti has quit IRC | 18:42 | |
| *** david-lyle has quit IRC | 18:42 | |
| *** ociuhandu has joined #openstack-neutron | 18:44 | |
| *** fzdarsky is now known as fzdarsky|afk | 18:45 | |
| *** yfried has quit IRC | 18:45 | |
| *** yfried has joined #openstack-neutron | 18:47 | |
| jschwarz | amuller, https://review.openstack.org/348224 , https://review.openstack.org/348215, https://review.openstack.org/265680 | 18:47 |
| *** buttercup has quit IRC | 18:49 | |
| *** nlahouti has joined #openstack-neutron | 18:51 | |
| *** nlahouti1 has quit IRC | 18:52 | |
| *** elopez_ has quit IRC | 18:53 | |
| openstackgerrit | Michael Bayer proposed openstack/neutron: Transition to new oslo_db test fixtures https://review.openstack.org/347609 | 18:54 |
| *** baoli has joined #openstack-neutron | 18:54 | |
| *** vijaykc4 has quit IRC | 18:56 | |
| *** nlahouti has quit IRC | 18:58 | |
| *** nlahouti1 has joined #openstack-neutron | 18:58 | |
| *** vijaykc4 has joined #openstack-neutron | 18:59 | |
| openstackgerrit | Anindita Das proposed openstack/neutron: Refactoring config options for l2 agent ext opts https://review.openstack.org/348513 | 18:59 |
| *** mickeys has quit IRC | 18:59 | |
| *** obondarev has joined #openstack-neutron | 18:59 | |
| *** mickeys has joined #openstack-neutron | 19:00 | |
| *** vijaykc4 has quit IRC | 19:00 | |
| *** akshai has quit IRC | 19:02 | |
| *** fifieldt has quit IRC | 19:02 | |
| *** mickeys has quit IRC | 19:04 | |
| *** buttercup has joined #openstack-neutron | 19:05 | |
| *** vijaykc4 has joined #openstack-neutron | 19:07 | |
| *** gouthamr has quit IRC | 19:10 | |
| *** fzdarsky|afk has quit IRC | 19:11 | |
| *** numans has quit IRC | 19:12 | |
| *** Sukhdev has quit IRC | 19:17 | |
| *** fifieldt has joined #openstack-neutron | 19:18 | |
| *** elo has quit IRC | 19:18 | |
| *** eric_lopez has joined #openstack-neutron | 19:18 | |
| *** lori has quit IRC | 19:20 | |
| *** abregman has joined #openstack-neutron | 19:23 | |
| openstackgerrit | Rawlin Peters proposed openstack/neutron: Pass bridge_name in OVS port's vif_details https://review.openstack.org/343816 | 19:25 |
| *** fzdarsky|afk has joined #openstack-neutron | 19:25 | |
| *** itzikb has quit IRC | 19:26 | |
| *** fifieldt has quit IRC | 19:27 | |
| *** matrohon has joined #openstack-neutron | 19:30 | |
| *** tonytan4ever has joined #openstack-neutron | 19:30 | |
| *** yuelongguang has quit IRC | 19:31 | |
| *** yuelongguang has joined #openstack-neutron | 19:31 | |
| *** gvrangan has joined #openstack-neutron | 19:32 | |
| *** singhj has joined #openstack-neutron | 19:35 | |
| *** yfried has quit IRC | 19:36 | |
| openstackgerrit | Aradhana Singh proposed openstack/neutron: DHCP Auto Scheduling for routed provider networks https://review.openstack.org/333716 | 19:36 |
| *** fifieldt has joined #openstack-neutron | 19:37 | |
| amuller | armax: https://review.openstack.org/#/c/337064/ | 19:40 |
| *** mvk has quit IRC | 19:40 | |
| *** dane_leblanc has quit IRC | 19:40 | |
| *** claudiub|2 has quit IRC | 19:41 | |
| *** obondarev has quit IRC | 19:43 | |
| openstackgerrit | Anindita Das proposed openstack/neutron: Refactoring config options for extension opts https://review.openstack.org/345486 | 19:43 |
| *** jckasper has quit IRC | 19:45 | |
| *** jckasper has joined #openstack-neutron | 19:45 | |
| *** tflynn has joined #openstack-neutron | 19:46 | |
| *** diga has quit IRC | 19:47 | |
| *** eric_lopez has quit IRC | 19:49 | |
| *** tflynn has quit IRC | 19:50 | |
| *** jckasper has quit IRC | 19:50 | |
| *** elopez_ has joined #openstack-neutron | 19:50 | |
| *** dane_leblanc has joined #openstack-neutron | 19:51 | |
| *** jckasper has joined #openstack-neutron | 19:51 | |
| *** elopez_ has quit IRC | 19:54 | |
| *** elopez_ has joined #openstack-neutron | 19:55 | |
| *** lori has joined #openstack-neutron | 19:55 | |
| *** jckasper has quit IRC | 19:56 | |
| *** obondarev has joined #openstack-neutron | 19:57 | |
| openstackgerrit | boden proposed openstack/neutron-lib: Add Neutron context module and some policy methods https://review.openstack.org/303867 | 19:59 |
| *** rossella_s has quit IRC | 19:59 | |
| *** rossella_s has joined #openstack-neutron | 20:00 | |
| *** jprovazn has quit IRC | 20:01 | |
| *** jamesdenton has quit IRC | 20:02 | |
| *** baoli has quit IRC | 20:04 | |
| *** obondarev has quit IRC | 20:05 | |
| *** jckasper has joined #openstack-neutron | 20:07 | |
| *** janzian has quit IRC | 20:11 | |
| *** tbachman has quit IRC | 20:11 | |
| *** matrohon has quit IRC | 20:22 | |
| *** tbachman has joined #openstack-neutron | 20:22 | |
| *** gvrangan has quit IRC | 20:22 | |
| *** janzian has joined #openstack-neutron | 20:22 | |
| *** eilert has quit IRC | 20:22 | |
| *** slaweq has joined #openstack-neutron | 20:23 | |
| *** sindhu has quit IRC | 20:23 | |
| *** dasanind has quit IRC | 20:24 | |
| *** vijaykc4 has quit IRC | 20:25 | |
| *** slaweq has quit IRC | 20:25 | |
| *** fragatina has joined #openstack-neutron | 20:27 | |
| *** zkassab is now known as z_kassab | 20:29 | |
| *** z_kassab is now known as zkassab | 20:29 | |
| openstackgerrit | Aradhana Singh proposed openstack/neutron: DHCP Auto Scheduling for routed provider networks https://review.openstack.org/333716 | 20:29 |
| *** lori has quit IRC | 20:30 | |
| *** sindhu has joined #openstack-neutron | 20:30 | |
| *** dasanind has joined #openstack-neutron | 20:30 | |
| *** lori has joined #openstack-neutron | 20:30 | |
| *** lori has quit IRC | 20:32 | |
| *** lori has joined #openstack-neutron | 20:33 | |
| *** zkassab has quit IRC | 20:34 | |
| *** Swami has quit IRC | 20:34 | |
| *** andymaier has joined #openstack-neutron | 20:36 | |
| *** Swami has joined #openstack-neutron | 20:37 | |
| *** mickeys has joined #openstack-neutron | 20:44 | |
| *** emagana has joined #openstack-neutron | 20:45 | |
| *** mfranc213_ has quit IRC | 20:47 | |
| *** eilert has joined #openstack-neutron | 20:47 | |
| *** singhj has quit IRC | 20:48 | |
| *** mfranc213 has joined #openstack-neutron | 20:49 | |
| mfranc213 | hello ihrachys: i hope to get your thoughts on https://review.openstack.org/#/c/339246/10/neutron/agent/l3/agent.py@433 | 20:49 |
| *** emagana has quit IRC | 20:49 | |
| *** singhj has joined #openstack-neutron | 20:50 | |
| *** matrohon has joined #openstack-neutron | 20:53 | |
| *** fzdarsky|afk has quit IRC | 20:54 | |
| *** cleong has quit IRC | 20:54 | |
| *** iranzo has joined #openstack-neutron | 20:56 | |
| *** jamielennox|away is now known as jamielennox | 20:57 | |
| *** marst has quit IRC | 21:02 | |
| *** rtheis has quit IRC | 21:02 | |
| *** marst has joined #openstack-neutron | 21:03 | |
| *** elo has joined #openstack-neutron | 21:04 | |
| *** marst has quit IRC | 21:05 | |
| *** tbachman_ has joined #openstack-neutron | 21:06 | |
| *** tbachman has quit IRC | 21:08 | |
| *** tbachman_ is now known as tbachman | 21:08 | |
| *** yamamoto has joined #openstack-neutron | 21:09 | |
| openstackgerrit | Richard Theis proposed openstack/neutron: Support L3 plugins without agent https://review.openstack.org/348558 | 21:09 |
| *** iranzo has quit IRC | 21:12 | |
| *** matrohon has quit IRC | 21:13 | |
| openstackgerrit | Manjeet Singh Bhatia proposed openstack/neutron: Refactor Router Db Models https://review.openstack.org/348562 | 21:14 |
| *** buttercup has quit IRC | 21:16 | |
| *** singhj has quit IRC | 21:16 | |
| *** singhj has joined #openstack-neutron | 21:18 | |
| *** marst has joined #openstack-neutron | 21:19 | |
| *** gvrangan has joined #openstack-neutron | 21:21 | |
| *** yamamoto has quit IRC | 21:21 | |
| *** abhiraut1 has joined #openstack-neutron | 21:24 | |
| openstackgerrit | Abhishek Raut proposed openstack/python-neutronclient: Add trunk commands to openstackclient https://review.openstack.org/340624 | 21:25 |
| *** andymaier has quit IRC | 21:26 | |
| openstackgerrit | Anindita Das proposed openstack/neutron: Refactoring config options for services opts https://review.openstack.org/347044 | 21:26 |
| *** abhiraut has quit IRC | 21:27 | |
| *** nlahouti has joined #openstack-neutron | 21:28 | |
| *** nlahouti1 has quit IRC | 21:28 | |
| *** abhiraut has joined #openstack-neutron | 21:29 | |
| *** dane_leblanc has quit IRC | 21:29 | |
| *** abhiraut1 has quit IRC | 21:31 | |
| *** kbringard has joined #openstack-neutron | 21:32 | |
| *** nlahouti1 has joined #openstack-neutron | 21:32 | |
| *** nlahouti has quit IRC | 21:32 | |
| *** thorst has quit IRC | 21:33 | |
| *** thorst has joined #openstack-neutron | 21:34 | |
| *** baoli has joined #openstack-neutron | 21:34 | |
| *** dasanind has quit IRC | 21:36 | |
| *** nlahouti1 has quit IRC | 21:36 | |
| *** nlahouti has joined #openstack-neutron | 21:36 | |
| *** sindhu has quit IRC | 21:37 | |
| *** thorst has quit IRC | 21:38 | |
| *** edmondsw has quit IRC | 21:46 | |
| *** rahuls_ has quit IRC | 21:46 | |
| *** iranzo has joined #openstack-neutron | 21:47 | |
| openstackgerrit | Pablo Iranzo Gómez proposed openstack/neutron-lib: Enhance valid_values validations to check that valid_values has method __contains__ https://review.openstack.org/343799 | 21:49 |
| *** nlahouti has quit IRC | 21:50 | |
| *** nlahouti has joined #openstack-neutron | 21:51 | |
| *** gouthamr has joined #openstack-neutron | 21:54 | |
| *** nlahouti1 has joined #openstack-neutron | 21:55 | |
| *** iranzo has quit IRC | 21:55 | |
| *** nlahouti has quit IRC | 21:55 | |
| *** nlahouti1 has quit IRC | 21:58 | |
| *** nlahouti has joined #openstack-neutron | 21:58 | |
| *** nlahouti1 has joined #openstack-neutron | 22:01 | |
| *** nlahouti has quit IRC | 22:01 | |
| *** nlahouti1 has quit IRC | 22:01 | |
| *** nlahouti2 has joined #openstack-neutron | 22:01 | |
| *** charliekang has quit IRC | 22:05 | |
| *** marst has quit IRC | 22:09 | |
| *** marst has joined #openstack-neutron | 22:09 | |
| *** wolverineav has joined #openstack-neutron | 22:09 | |
| *** tonytan4ever has quit IRC | 22:10 | |
| *** wolverineav has quit IRC | 22:13 | |
| *** gouthamr_ has joined #openstack-neutron | 22:16 | |
| openstackgerrit | Aradhana Singh proposed openstack/neutron: Refactoring config options of l3 agent keepalived https://review.openstack.org/338559 | 22:19 |
| *** gouthamr has quit IRC | 22:19 | |
| *** abregman has quit IRC | 22:20 | |
| *** yamamoto has joined #openstack-neutron | 22:22 | |
| openstackgerrit | Merged openstack/neutron: Add some negative policy router interface tests https://review.openstack.org/348071 | 22:22 |
| *** boden has quit IRC | 22:23 | |
| *** yamamoto has quit IRC | 22:27 | |
| openstackgerrit | Michael Bayer proposed openstack/neutron: Transition to new oslo_db test fixtures https://review.openstack.org/347609 | 22:29 |
| *** krtaylor has quit IRC | 22:31 | |
| *** amuller has quit IRC | 22:34 | |
| *** wolverineav has joined #openstack-neutron | 22:37 | |
| *** mickeys has quit IRC | 22:42 | |
| *** wolverineav has quit IRC | 22:42 | |
| *** wolverineav has joined #openstack-neutron | 22:44 | |
| *** yamahata has quit IRC | 22:47 | |
| *** wolverineav has quit IRC | 22:51 | |
| *** sindhu has joined #openstack-neutron | 22:58 | |
| *** claudiub|2 has joined #openstack-neutron | 22:59 | |
| *** yamahata has joined #openstack-neutron | 23:00 | |
| *** nlahouti2 has quit IRC | 23:01 | |
| *** nlahouti has joined #openstack-neutron | 23:01 | |
| *** wolverineav has joined #openstack-neutron | 23:02 | |
| *** fnaval has joined #openstack-neutron | 23:02 | |
| *** fnaval has quit IRC | 23:02 | |
| *** liuyulong has joined #openstack-neutron | 23:03 | |
| *** mickeys has joined #openstack-neutron | 23:03 | |
| *** fnaval has joined #openstack-neutron | 23:03 | |
| *** pradk has quit IRC | 23:04 | |
| *** jamesdenton has joined #openstack-neutron | 23:06 | |
| *** jamesdenton has quit IRC | 23:06 | |
| *** wolverineav has quit IRC | 23:06 | |
| *** mickeys has quit IRC | 23:07 | |
| *** mvk has joined #openstack-neutron | 23:07 | |
| *** nlahouti has quit IRC | 23:07 | |
| *** nlahouti has joined #openstack-neutron | 23:07 | |
| *** tpsilva has quit IRC | 23:08 | |
| *** a_ta has quit IRC | 23:10 | |
| *** a_ta has joined #openstack-neutron | 23:11 | |
| *** tonytan4ever has joined #openstack-neutron | 23:11 | |
| *** nlahouti has quit IRC | 23:11 | |
| *** sdague has quit IRC | 23:11 | |
| *** nlahouti has joined #openstack-neutron | 23:11 | |
| *** itzikb has joined #openstack-neutron | 23:11 | |
| *** kbringard has quit IRC | 23:13 | |
| *** a_ta has quit IRC | 23:15 | |
| *** tonytan4ever has quit IRC | 23:16 | |
| *** emagana has joined #openstack-neutron | 23:16 | |
| *** andymaier has joined #openstack-neutron | 23:18 | |
| *** emagana has quit IRC | 23:20 | |
| openstackgerrit | LIU Yulong proposed openstack/neutron: Filter HA router without HA port bindings after race conditions https://review.openstack.org/265680 | 23:21 |
| *** nlahouti has quit IRC | 23:22 | |
| *** jamesdenton has joined #openstack-neutron | 23:22 | |
| *** nlahouti has joined #openstack-neutron | 23:22 | |
| *** tonytan4ever has joined #openstack-neutron | 23:23 | |
| *** Leom has quit IRC | 23:23 | |
| *** yamamoto has joined #openstack-neutron | 23:24 | |
| *** jamesden_ has joined #openstack-neutron | 23:24 | |
| *** tonytan_brb has joined #openstack-neutron | 23:24 | |
| *** wolverineav has joined #openstack-neutron | 23:24 | |
| *** mickeys has joined #openstack-neutron | 23:26 | |
| *** jamesdenton has quit IRC | 23:28 | |
| *** tonytan4ever has quit IRC | 23:28 | |
| *** yamamoto has quit IRC | 23:29 | |
| *** wolverineav has quit IRC | 23:29 | |
| *** nlahouti has quit IRC | 23:33 | |
| *** nlahouti has joined #openstack-neutron | 23:33 | |
| *** elopez_ has quit IRC | 23:34 | |
| *** wolverineav has joined #openstack-neutron | 23:34 | |
| *** itzikb has quit IRC | 23:35 | |
| *** eilert has quit IRC | 23:36 | |
| *** singhj has quit IRC | 23:36 | |
| *** sdague has joined #openstack-neutron | 23:38 | |
| *** djan has joined #openstack-neutron | 23:38 | |
| *** singhj has joined #openstack-neutron | 23:39 | |
| *** wolverineav has quit IRC | 23:39 | |
| *** nlahouti1 has joined #openstack-neutron | 23:41 | |
| *** nlahouti has quit IRC | 23:41 | |
| *** sdague has quit IRC | 23:43 | |
| *** nlahouti1 has quit IRC | 23:44 | |
| *** nlahouti has joined #openstack-neutron | 23:44 | |
| *** liuyulong has quit IRC | 23:45 | |
| *** hoangcx2 has joined #openstack-neutron | 23:46 | |
| *** nlahouti1 has joined #openstack-neutron | 23:47 | |
| *** nlahouti has quit IRC | 23:47 | |
| *** hoangcx has quit IRC | 23:47 | |
| openstackgerrit | Merged openstack/neutron: Add flavor/service provider support to routers https://review.openstack.org/268941 | 23:48 |
| *** wolverineav has joined #openstack-neutron | 23:50 | |
| *** nlahouti1 has quit IRC | 23:51 | |
| *** nlahouti has joined #openstack-neutron | 23:51 | |
| *** tonytan_brb has quit IRC | 23:52 | |
| *** itlinux has joined #openstack-neutron | 23:55 | |
| *** nlahouti has quit IRC | 23:56 | |
| *** nlahouti1 has joined #openstack-neutron | 23:56 | |
| *** yb has joined #openstack-neutron | 23:56 | |
| *** nlahouti1 has quit IRC | 23:58 | |
| *** wolverineav has quit IRC | 23:58 | |
| *** nlahouti has joined #openstack-neutron | 23:58 | |
| *** wolverineav has joined #openstack-neutron | 23:58 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!