| opendevreview | Elvira García Ruiz proposed openstack/neutron master: Consider logging options when using OVNdbsync https://review.opendev.org/c/openstack/neutron/+/948783 | 08:02 |
|---|---|---|
| opendevreview | Martin André proposed openstack/neutron-lib master: Fix documentation for gateway_ip in subnet update https://review.opendev.org/c/openstack/neutron-lib/+/952428 | 08:40 |
| opendevreview | Elvira García Ruiz proposed openstack/neutron master: Consider logging options when using OVNdbsync https://review.opendev.org/c/openstack/neutron/+/948783 | 10:02 |
| *** qwebirc70632 is now known as agraeb | 12:11 | |
| agraeb | Hi there, is this the right place to ask questions about DevStack and Neutron? | 12:14 |
| ralonsoh | agraeb, maybe we can solve the issue in this channel | 12:49 |
| ralonsoh | well, issue or question | 12:49 |
| opendevreview | Rodolfo Alonso proposed openstack/neutron master: Move the ``ProcessMonitor`` initialization to the ``Service.start`` phase https://review.opendev.org/c/openstack/neutron/+/952450 | 12:49 |
| agraeb | I have a multi node DevStack setup (one control plane and one compute) with OVN. The compute node didn't create the vxlan-* interface on the tunnel bridge. Meanwhile I was able to solve that by restarting the OpenVSwitch agent on the control plane node. I suppose by doint so, the existance of the network agent on the control plane node was accounced to the agent on the compute node. | 13:00 |
| ralonsoh | slaweq, haleyb if you have 1 min: | 13:37 |
| ralonsoh | * https://review.opendev.org/c/openstack/neutron/+/952193 | 13:37 |
| ralonsoh | * https://review.opendev.org/c/openstack/neutron/+/952209 | 13:37 |
| ralonsoh | thanks! | 13:38 |
| ralonsoh | (both have testing patches on top) | 13:38 |
| opendevreview | Rodolfo Alonso proposed openstack/neutron master: Allow empty gateway IP in subnets from subnet pools https://review.opendev.org/c/openstack/neutron/+/952335 | 13:39 |
| haleyb | ralonsoh: ack, will look | 14:19 |
| ralonsoh | thanks!@ | 14:19 |
| ralonsoh | haleyb, ah, we found an issue with L3 and the eventlet removal | 14:19 |
| haleyb | ralonsoh: also, i created https://review.opendev.org/c/openstack/neutron/+/952399 yesterday, but it's clear that removing eventlet from the metadata agent broke this, and we'll need to figure out something | 14:19 |
| ralonsoh | --> https://review.opendev.org/c/openstack/neutron/+/952450 | 14:19 |
| haleyb | ralonsoh: seems there's more than one eventlet removal issue :( | 14:20 |
| ralonsoh | haleyb, let me check your patch | 14:20 |
| haleyb | ralonsoh: read my comment in the patch as well, which i think describes it | 14:20 |
| haleyb | i.e. i don't know why that Looping thread is not running | 14:21 |
| ralonsoh | haleyb, ok so metadata is not reporting the status | 14:22 |
| ralonsoh | (no problem! we can harcode it!!!) | 14:22 |
| ralonsoh | let me restart my ml2/ovs env and deploy it | 14:22 |
| haleyb | ralonsoh: nope, and i was finally able to reproduce it. putting a self._report_state() call there showed it happy for 30 seconds :) | 14:22 |
| ralonsoh | haleyb, yes but then we don't send any further update, right? | 14:23 |
| haleyb | nope. and i never see that FixedIntervalLoopingCall() do anything | 14:24 |
| ralonsoh | haleyb, ok, let me check that this evening or tomorrow morning, I need to warp up some patches | 14:25 |
| haleyb | ralonsoh: np, my guess is the ovn-metadata-agent would be just as broken as they share a lot of code. maybe looking at your l3-agent patch will help my brain | 14:29 |
| ralonsoh | the l3 patch is very simple | 14:30 |
| ralonsoh | we found that in a tobiko job | 14:30 |
| ralonsoh | actually I need to send a testing patch for tobiko now | 14:30 |
| opendevreview | Rodolfo Alonso proposed x/whitebox-neutron-tempest-plugin master: Adapt devstack WSGI check for deprecated configuration https://review.opendev.org/c/x/whitebox-neutron-tempest-plugin/+/951264 | 14:40 |
| haleyb | ralonsoh: after looking at https://review.opendev.org/c/openstack/neutron/+/938404 there might be some similarites to the metadata-agent code - i.e. both use ThreadingUnixStreamServer() now so might need something similar. i'll hack at it if i have time today | 14:48 |
| opendevreview | Merged openstack/neutron-lib master: Fix documentation for gateway_ip in subnet update https://review.opendev.org/c/openstack/neutron-lib/+/952428 | 14:48 |
| ralonsoh | haleyb, I'll check that too | 14:48 |
| haleyb | that run()/wait() change catches my eye at least | 14:49 |
| haleyb | ralonsoh: the other thing i noticed yesterday is we are hitting the 1-hour time limit in the unit test jobs more, I was going to send a patch to increase the limit from 3600 to 4800 | 14:58 |
| ralonsoh | haleyb, yes, that was solved here | 14:58 |
| ralonsoh | no no | 14:58 |
| ralonsoh | one sec | 14:58 |
| ralonsoh | haleyb, https://review.opendev.org/c/openstack/neutron/+/952220 this should fix that | 14:58 |
| ralonsoh | check the pyx jobs there | 14:59 |
| ralonsoh | 18mins, 23mins | 14:59 |
| haleyb | ah, that's much better, i was only looking at the past few days and saw everything close to 1 hour | 14:59 |
| ralonsoh | haleyb, btw, I've reproduced the L3 agent dead | 15:00 |
| ralonsoh | but | 15:00 |
| ralonsoh | it is flapping | 15:00 |
| ralonsoh | so there is something weird there, I'll check better tomorrow morning | 15:00 |
| haleyb | ralonsoh: ack, thanks! damn eventlet | 15:01 |
| ralonsoh | yeah... | 15:01 |
| ralonsoh | haleyb, dammit | 15:07 |
| ralonsoh | we are calling the oslo.service eventlet backend, not the threading | 15:07 |
| ralonsoh | we need to push a patch with something like this | 15:07 |
| ralonsoh | https://review.opendev.org/c/openstack/neutron/+/938404/26/neutron/cmd/agents/l3.py | 15:08 |
| ralonsoh | actually for all agents not patched (metadata, ovn metadta and ovn agent) | 15:08 |
| ralonsoh | all 3 are not patched correctly | 15:09 |
| ralonsoh | haleyb, and now is working... heheheh | 15:09 |
| ralonsoh | please, if you can do this, send this patch, I need to leave in 5 mins | 15:10 |
| ralonsoh | thanks! | 15:10 |
| haleyb | ralonsoh: good find! i can try the metadata-agent locally if my stack is still up. we do need to put back the call that was accidentally removed but that's easy :) | 15:10 |
| ralonsoh | yes, this call you are adding is mandatory | 15:11 |
| haleyb | ralonsoh: tested locally and it fixed the issue, all :-) and can see _report_state() looping now | 15:18 |
| cardoe | haleyb: Since this patch conflicts with your ovs-agent change, I was wondering if you'd be willing to review it https://review.opendev.org/c/openstack/neutron/+/945497 | 16:32 |
| cardoe | We're trying to get that in for https://review.opendev.org/c/openstack/networking-baremetal/+/945500 | 16:32 |
| cardoe | Happy to get them to rebase on top of yours | 16:33 |
| haleyb | cardoe: which ovs-agent change? sorry the only merge conflict it shows is https://review.opendev.org/c/openstack/neutron/+/868151 | 16:36 |
| opendevreview | Brian Haley proposed openstack/neutron master: Add state reporting back to metadata agents https://review.opendev.org/c/openstack/neutron/+/952399 | 16:36 |
| opendevreview | Brian Haley proposed openstack/neutron master: Add state reporting back to metadata agents https://review.opendev.org/c/openstack/neutron/+/952399 | 16:41 |
| cardoe | I thought what you were talking about was 868151. my mistake. | 16:45 |
| haleyb | cardoe: no, it was some eventlet-related bugs, np | 16:48 |
| opendevreview | Merged openstack/networking-bgpvpn master: Set service type when registering provider config https://review.opendev.org/c/openstack/networking-bgpvpn/+/950212 | 16:54 |
| *** jlibosva is now known as Guest18037 | 19:13 | |
| opendevreview | Merged x/whitebox-neutron-tempest-plugin master: Adapt devstack WSGI check for deprecated configuration https://review.opendev.org/c/x/whitebox-neutron-tempest-plugin/+/951264 | 19:22 |
| *** jlibosva is now known as Guest18040 | 19:44 | |
| haleyb | has anyone seen this error running 'tox -e pep8' on a freshly cloned neutron repo? "NameError: name 'network_id' is not defined" - it isn't giving me any clues on what file it's coming from unfortunately no matter what options I use | 20:57 |
| haleyb | only 211 occurences of network_id in the tree according to flake8 :( | 21:00 |
| haleyb | by process of elimination i've figured out it's in neutron/agent/metadata/agent.py | 21:12 |
| opendevreview | Brian Haley proposed openstack/neutron master: Fix bad i18n and fstrings interaction with flake8 https://review.opendev.org/c/openstack/neutron/+/952524 | 21:43 |
Generated by irclog2html.py 4.0.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!