| opendevreview | liuyulong proposed openstack/neutron master: Refactor for meter ID Generator https://review.opendev.org/c/openstack/neutron/+/860765 | 01:44 |
|---|---|---|
| opendevreview | liuyulong proposed openstack/neutron master: Refactor for ovs qos driver meter limit features https://review.opendev.org/c/openstack/neutron/+/860766 | 01:44 |
| opendevreview | liuyulong proposed openstack/neutron master: Add meter bandwidth bandwidth support https://review.opendev.org/c/openstack/neutron/+/860767 | 01:44 |
| opendevreview | liuyulong proposed openstack/neutron-lib master: Add flows tables for istributed metadata datapath https://review.opendev.org/c/openstack/neutron-lib/+/861810 | 01:54 |
| opendevreview | liuyulong proposed openstack/neutron-lib master: Add flows tables for distributed metadata datapath https://review.opendev.org/c/openstack/neutron-lib/+/861810 | 07:29 |
| opendevreview | Merged openstack/neutron master: Do not keep gateway port when notifying for router update https://review.opendev.org/c/openstack/neutron/+/861322 | 07:45 |
| opendevreview | Michal Arbet proposed openstack/neutron master: Remove xenapi from neutron ml2 config opts https://review.opendev.org/c/openstack/neutron/+/861828 | 07:57 |
| opendevreview | liuyulong proposed openstack/neutron master: Refactor metedata signature function https://review.opendev.org/c/openstack/neutron/+/861829 | 08:05 |
| opendevreview | Dr. Jens Harbott proposed openstack/neutron master: Remove xenapi entrypoint https://review.opendev.org/c/openstack/neutron/+/861852 | 08:30 |
| zigo | Hi there! | 08:55 |
| zigo | Rebuilding neutron/zed in Unstable leads to 200+ unit test failures for me, all like these 2: | 08:55 |
| zigo | https://paste.opendev.org/show/bBSH1STe0Yql8LfqeWcx/ | 08:55 |
| zigo | Can someone help me find the issue? | 08:55 |
| frickler | zigo: how do you run your tests? iirc you're not using tox? also which version of oslo-config is this? | 08:58 |
| zigo | frickler: I run my tests running stestr, with whatever the version of the lib that is in Sid. | 08:59 |
| zigo | Tox downloads depends from the internet, which is forbiden during a package build. | 08:59 |
| frickler | yes, I'm just wondering why the issue appears in your builds but not in upstream CI | 09:00 |
| frickler | oh, wait, I can reproduce this if I am running a single isolated test with tox | 09:01 |
| zigo | Oh... | 09:02 |
| zigo | Yet another test order thingy?!? | 09:02 |
| frickler | tox -e py3 -- neutron.tests.unit.plugins.ml2.drivers.ovn.mech_driver.test_mech_driver.TestOVNMechanismDriverSecurityGroup | 09:03 |
| frickler | this is failing | 09:03 |
| frickler | tox -e py3 -- neutron.tests.unit.plugins.ml2.drivers.ovn.mech_driver.test_mech_driver | 09:03 |
| frickler | works fine | 09:03 |
| zigo | I probably can just blacklist all of neutron.tests.unit.plugins.ml2.drivers.ovn.mech_driver then. | 09:03 |
| frickler | I'll leave it to neutron devs to dig into it | 09:03 |
| frickler | zigo: are all your failures in that module? | 09:04 |
| zigo | Let me check. | 09:04 |
| zigo | Nop. | 09:05 |
| zigo | Some in neutron.tests.unit.services.ovn_l3.test_plugin | 09:05 |
| frickler | but I get the same effect there, fails if I run tox with just the single test | 09:06 |
| zigo | So, all in neutron.tests.unit.plugins.ml2.drivers.ovn.mech_driver.test_mech_driver.TestOVNMechanismDriverSecurityGroup or in neutron.tests.unit.services.ovn_l3.test_plugin.OVNL3ExtrarouteTests | 09:06 |
| frickler | zigo: I guess it will be helpful if you could create a bug report for tracking this | 09:08 |
| zigo | Yeah. Let me finish rebuilding Neutron with kevko's patch first. | 09:12 |
| zigo | frickler: https://bugs.launchpad.net/neutron/+bug/1993502 | 09:23 |
| frickler | zigo: thx | 09:25 |
| opendevreview | Rodolfo Alonso proposed openstack/neutron master: Add new oslo master CI jobs to the periodic queue https://review.opendev.org/c/openstack/neutron/+/861859 | 09:30 |
| zigo | This still fails with kevko's patch, even when skipping the failing unit tests... :/ | 09:31 |
| ralonsoh | zigo, but the error is clear in the logs | 09:33 |
| ralonsoh | by default, you should execute a module, not a class | 09:33 |
| ralonsoh | (by default) | 09:33 |
| ralonsoh | in any case, if executing a single class returns this error | 09:33 |
| ralonsoh | this is because this class is not explicitly importing this config var | 09:33 |
| frickler | ralonsoh: "tox -e py3 -- neutron.tests.unit.services.ovn_l3" is failing for me, too | 09:38 |
| frickler | and IMHO each test should be able to run on it's own | 09:38 |
| ralonsoh | frickler, ok, I'll check it | 09:51 |
| opendevreview | Slawek Kaplonski proposed openstack/neutron master: [WIP] Switch fullstack/functional fips jobs to 9-stream https://review.opendev.org/c/openstack/neutron/+/843245 | 09:58 |
| opendevreview | Merged openstack/neutron stable/xena: Disable in-band management for bridges before setting up controllers https://review.opendev.org/c/openstack/neutron/+/861624 | 09:58 |
| ralonsoh | ykarel, hi! how to I check the "experimental" queue in https://review.opendev.org/c/openstack/neutron/+/861859 ? | 10:12 |
| ralonsoh | no no | 10:12 |
| ralonsoh | not the experimental, the periodic one | 10:12 |
| opendevreview | Merged openstack/neutron stable/train: Disable in-band management for bridges before setting up controllers https://review.opendev.org/c/openstack/neutron/+/861707 | 10:17 |
| opendevreview | Merged openstack/neutron stable/yoga: Disable in-band management for bridges before setting up controllers https://review.opendev.org/c/openstack/neutron/+/861623 | 10:17 |
| opendevreview | Merged openstack/neutron master: Remove xenapi from neutron ml2 config opts https://review.opendev.org/c/openstack/neutron/+/861828 | 10:17 |
| ralonsoh | slaweq, do you know how to trigger the periodic queue in a patch? | 10:25 |
| ralonsoh | check periodic ? | 10:25 |
| opendevreview | Rodolfo Alonso proposed openstack/neutron master: Load the required configuration options in the UT classes https://review.opendev.org/c/openstack/neutron/+/861869 | 10:29 |
| opendevreview | Rodolfo Alonso proposed openstack/neutron master: Add new oslo master CI jobs to the periodic queue https://review.opendev.org/c/openstack/neutron/+/861859 | 10:30 |
| frickler | ralonsoh: there is no trigger. you can add a job to the check queue in a test patch instead | 10:54 |
| ralonsoh | frickler, I added those new tests to the experimental queue | 10:54 |
| ralonsoh | new jobs* | 10:55 |
| frickler | ralonsoh: I see those running in the experimental queue, so seems to be fine. "check experimental" is the right trigger for that | 10:57 |
| ralonsoh | thanks! | 10:57 |
| frickler | just note that in contrast to "recheck", no extra comment is allowed with it | 10:57 |
| ykarel | ralonsoh, just back from lunch, as frickler said no other trigger for periodic pipeline | 11:15 |
| ykarel | and we have those jobs in experimental, so running those will do it | 11:15 |
| ralonsoh | ykarel, yeah, I added those jobs to the experimental queue | 11:15 |
| *** sean-k-mooney1 is now known as sean-k-mooney | 11:31 | |
| ralonsoh | hi folks, meeting now in Mitaka room | 13:06 |
| ralonsoh | lajoskatona, we have functional and fullstack with puyroute2 master | 13:41 |
| ralonsoh | in periodic | 13:41 |
| lajoskatona | ralonsoh: ok, thanks | 13:41 |
| *** dasm|off is now known as dasm | 13:55 | |
| opendevreview | Pierre Libeau proposed openstack/neutron master: Add default policy on device_id param create_port https://review.opendev.org/c/openstack/neutron/+/861169 | 14:55 |
| haleyb | lucasagomes: i think the problem, at least on my system, is that I already have both directories in /var/run, so that 'ln -s' line puts a link in /var/run/ovn, it doesn't make a link of /var/run/ovn to /var/run/openvswitch | 14:57 |
| lucasagomes | haleyb, a-ha... I think it's ovs-vswitchd that create that directory... maybe it's running at the time the symlink is created | 14:58 |
| lucasagomes | that's why the dir is already there, maybe a fix would be to stop it prior to creating the symlink ? | 14:58 |
| lucasagomes | and we create the dirs ourselves in OVN | 14:58 |
| lucasagomes | in devstack* | 14:58 |
| lucasagomes | haleyb, or we could get rid of this symlink, and just change the code to look to the sockets in the right directories now as for OVN it will always be in /var/run/ovn now | 14:59 |
| haleyb | my system has both ovn-controller and ovs-vswitch running | 14:59 |
| lucasagomes | before it was depending whether the version of OVN was the old one that lived in the ovs repo, or the new one that was split out | 14:59 |
| lucasagomes | haleyb, right yeah that's probably the case | 14:59 |
| haleyb | lucasagomes: my fix was to just use the correct directories, i hadn't even seen the symlink code | 14:59 |
| lucasagomes | haleyb, r u using packages right ? | 15:00 |
| lucasagomes | I think ubuntu may start the services when u install the deb ? | 15:00 |
| lucasagomes | haleyb++ | 15:00 |
| haleyb | lucasagomes: yes, 22.04 packages, not building | 15:00 |
| lucasagomes | that's the better fix indeed | 15:00 |
| lucasagomes | just use the right directories | 15:00 |
| haleyb | lucasagomes: so remove the symlink? and see what breaks? :) | 15:00 |
| lucasagomes | haleyb, lol I guess :D | 15:01 |
| lucasagomes | haleyb, otherwise we can try to stop the services, create the dirs/symlink and then start then again | 15:01 |
| lucasagomes | that should do the trick | 15:01 |
| lucasagomes | them* | 15:01 |
| haleyb | lucasagomes: i don't know what's best, but probably don't need to stop/start if we're using the installed packages. Let me send a patch and make a neutron dependent one to see if it works | 15:03 |
| lucasagomes | haleyb, yeah, it's just that I think that ubuntu might start the services upon installing the deb | 15:06 |
| lucasagomes | haleyb, later on the code will start it anyway... I think in fact that we assumed that the services were stopped at that point of creating the symlinks | 15:06 |
| haleyb | lucasagomes: ah, i now see init_ovn/start_ovn which will cleanup any existing DBs and start. in my case they were already running. either way let me send a patch that also removes the symlink, doing a stack.sh now to test | 15:17 |
| lucasagomes | haleyb++ thanks for looking into it | 15:18 |
| haleyb | lucasagomes: i've had a local patch for months, figured i was a unicorn since it's not seen in the gate | 15:20 |
| lucasagomes | haleyb, haha yeah, to be fair, I don't recall seeing this either. Maybe different ubuntu versions ?! | 15:25 |
| *** ykarel is now known as ykarel|afk | 15:27 | |
| haleyb | lucasagomes: i think i saw it since i booted a new vm, but install ovs and ovn before doing anything, which creates /var/run/ovn. That messes up the devstack script. The gate uses a clean VM so never sees it since it creates the link, then installs packages. | 15:27 |
| lucasagomes | haleyb, that makes sense. Yeah I think that'sthe case indeed, installing the package will start the services in ubuntu... And we assumed there was no package there | 15:28 |
| lucasagomes | haleyb, looking fwd to try ur fix | 15:28 |
| haleyb | i hope it works, testing a new version now | 15:29 |
| gmann | slaweq: hi | 15:39 |
| *** ykarel|afk is now known as ykarel | 15:48 | |
| ralonsoh | gmann, he is away now | 15:49 |
| ralonsoh | lajoskatona, one qq about the OVN grenade jobs | 15:49 |
| ralonsoh | we have two jobs, "neutron-ovn-grenade-multinode-tick-tick" and "neutron-ovs-grenade-multinode-tick-tick" | 15:50 |
| ralonsoh | isn't that enough? | 15:50 |
| lajoskatona | ralonsoh: I think yes, with those we cover both the important backends, no? | 15:52 |
| ralonsoh | lajoskatona, maybe gmann can help us on this | 15:53 |
| gmann | ralonsoh: ohk, I just want to know if you are planning RBAC discussion for neutron? I could not find in etherpad. I would like to discuss about switching RBAC flag to true in this cycle? | 15:53 |
| ralonsoh | gmann, I'll check that tomorrow | 15:53 |
| ralonsoh | in any case, we have Friday for those meetings, if needed | 15:53 |
| gmann | ralonsoh: lajoskatona : *tick-tick jobs are skip level doing stable/yoga to master upgrade right? | 15:53 |
| ralonsoh | yes | 15:53 |
| ralonsoh | #link https://review.opendev.org/c/openstack/neutron/+/859991 | 15:54 |
| gmann | +1, those should be enough | 15:55 |
| ralonsoh | perfect, thanks a lot | 15:55 |
| lajoskatona | gmann, ralonsoh: exactly, from y to A | 15:55 |
| gmann | you can rename them to new name s/tick-tick/skip-level if you want :) | 15:55 |
| lajoskatona | cool than :-) | 15:55 |
| gmann | perfect | 15:55 |
| ralonsoh | we'll do it | 15:55 |
| gmann | thanks | 15:55 |
| gmann | ralonsoh: back to RBAC topic. I will be available tomorrow after 14 -15 UTC or friday 13-15 UTC. please let me know when you want to discuss. sorry for late request, i thought you have some discussion planned in etherpad and forget to ping early | 15:56 |
| ralonsoh | gmann, tomorrow impossible, Nova-neutron meetings. On friday we have the operator hours. Most probably Friday 14-15UTC will be fine | 15:57 |
| ralonsoh | I'll talk to slawek tomorrow | 15:57 |
| gmann | ralonsoh: perfect, friday works fine | 15:58 |
| gmann | thanks | 15:58 |
| gmann | other option can be TC tomorrow @17 UTC https://etherpad.opendev.org/p/tc-2023-1-ptg#L73 | 15:59 |
| ralonsoh | gmann, the agenda for tomorrow is crazy, not possible | 15:59 |
| ralonsoh | sorry | 15:59 |
| gmann | if you, slaweq and other neutron folks are ok with that time? | 15:59 |
| gmann | its after PTG slots @17 UTC | 15:59 |
| gmann | not in neutron slots. once you guys are done then we have TC slot continue | 16:00 |
| ralonsoh | I'll check that with slaweq tomorrow, a bit late but possible | 16:00 |
| ralonsoh | well, slaweq and other folks | 16:00 |
| gmann | ohk, right. i forget about TZ :) | 16:00 |
| gmann | let's chat tomorrow in morning to figure this out once slaweq online. | 16:02 |
| opendevreview | Rodolfo Alonso proposed openstack/neutron master: Change grenade job names suffix to "skip-level" https://review.opendev.org/c/openstack/neutron/+/861901 | 16:03 |
| frickler | zigo: the patch that ralonsoh made fixes the tox tests that I did, please check whether it fixes your scenario, too https://review.opendev.org/c/openstack/neutron/+/861869 | 16:17 |
| opendevreview | Merged openstack/neutron master: Load the required configuration options in the UT classes https://review.opendev.org/c/openstack/neutron/+/861869 | 16:36 |
| opendevreview | Brian Haley proposed openstack/neutron master: DNM: Test devstack change to ovn_agent code https://review.opendev.org/c/openstack/neutron/+/861916 | 18:18 |
| zigo | Thanks for the patch. I very much would welcome it to be backported to stable/zed ! | 20:38 |
| *** dasm is now known as dasm|off | 22:19 | |
| haleyb | zigo: if you're still around and click on the cherry-pick to get it to zed i could take a look and +2 | 23:01 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!