| gthiemonge | tkajinam: Hi Takashi, I think we have to backport https://review.opendev.org/c/openstack/octavia/+/903674 , it started to fail on stable branches too | 08:54 |
|---|---|---|
| gthiemonge | tkajinam: do you think it is fine to unpin flake8-import-order on stable branches? or do we need to take another approach? | 08:54 |
| tkajinam | gthiemonge, There would be two approaches. Backporting that fix to stable branch, which I think is row risk, or pin flake8 to < 6.0.0 in stable branch according to https://github.com/PyCQA/flake8-import-order/issues/189 | 09:10 |
| tkajinam | afaik flake8 is not capped by global requirements and that's why the issue is affecting stable branches | 09:11 |
| tkajinam | it seems the other repos such as neutron are not hitting this problem because flake8 version is capped by hacking. | 09:11 |
| tkajinam | Collecting flake8<3.8.0,>=3.6.0 (from hacking<3.1.0,>=3.0.1->-r /home/zuul/src/opendev.org/openstack/neutron-lib/test-requirements.txt (line 5)) | 09:12 |
| tkajinam | this is what I've found in pep8 job for neutron-lib | 09:12 |
| gthiemonge | tkajinam: thanks, maybe caping flake8 is the correct approach here, I'm going to check if it fixes the errors | 09:13 |
| tkajinam | gthiemonge, you can probably cap hacking then it should cap flake8 | 09:14 |
| tkajinam | or cap flake8 which is more direct approach | 09:15 |
| gthiemonge | tkajinam: ack | 09:16 |
| tkajinam | so... it turned out very old hacking versions are still pulled in multiple test-requirements. /me sighs | 09:16 |
| tkajinam | python-octaviaclient/test-requirements.txt:hacking>=3.0.1,<3.1.0 # Apache-2.0 | 09:17 |
| opendevreview | Takashi Kajinami proposed openstack/python-octaviaclient master: Bump hacking https://review.opendev.org/c/openstack/python-octaviaclient/+/905659 | 09:21 |
| opendevreview | Takashi Kajinami proposed openstack/octavia master: Cap hacking https://review.opendev.org/c/openstack/octavia/+/905660 | 09:22 |
| gthiemonge | tkajinam: thanks! | 09:23 |
| opendevreview | Takashi Kajinami proposed openstack/octavia stable/2023.2: Stable-only: Pin hacking to < 6.1.0 https://review.opendev.org/c/openstack/octavia/+/905662 | 09:27 |
| opendevreview | Gregory Thiemonge proposed openstack/octavia stable/2023.2: Cap hacking on stable branches https://review.opendev.org/c/openstack/octavia/+/905663 | 09:27 |
| gthiemonge | oh, almost same thing | 09:27 |
| gthiemonge | let's see if yours passes the CI maybe going from 3.0 to 6.0 will trigger new issues | 09:28 |
| tkajinam | gthiemonge, If I read the history in hacking. hacking 6.1.0 is the one which starts pulling flake8 >= 6 | 09:28 |
| tkajinam | so we can probably use 6.0.x, though yours would also work | 09:29 |
| tkajinam | I have no strong opinion and am file with either upper constraint, as long as it works | 09:30 |
| gthiemonge | I would rather not bump it on stable but maybe some other folks will have a different opinion | 09:32 |
| tkajinam | I'm ok with a lower bound. my concern is the fact that we haven't capped hacking so even old branches have been consumed new version recently and having too low cap might cause different problems. | 09:42 |
| tkajinam | anyway we can check the outcome in CI and use the one which fixes pep8 without different problems :-) | 09:42 |
| opendevreview | Takashi Kajinami proposed openstack/python-octaviaclient master: Bump hacking https://review.opendev.org/c/openstack/python-octaviaclient/+/905659 | 09:47 |
| opendevreview | Takashi Kajinami proposed openstack/octavia master: Cap hacking https://review.opendev.org/c/openstack/octavia/+/905660 | 09:49 |
| opendevreview | Takashi Kajinami proposed openstack/python-octaviaclient master: Bump hacking https://review.opendev.org/c/openstack/python-octaviaclient/+/905659 | 10:12 |
| opendevreview | Takashi Kajinami proposed openstack/octavia master: Cap hacking https://review.opendev.org/c/openstack/octavia/+/905660 | 10:40 |
| opendevreview | Tom Weininger proposed openstack/octavia-tempest-plugin master: WIP test rebase of below patch https://review.opendev.org/c/openstack/octavia-tempest-plugin/+/905683 | 11:33 |
| opendevreview | Tom Weininger proposed openstack/octavia-tempest-plugin master: WIP test rebase of below patch https://review.opendev.org/c/openstack/octavia-tempest-plugin/+/905683 | 11:58 |
| gthiemonge | tweining: johnsom: question for you guys, what do you prefer to fix a pep8 issue on stable branch | 13:42 |
| gthiemonge | - capping the current requirements: https://review.opendev.org/c/openstack/octavia/+/905663/1/test-requirements.txt | 13:42 |
| gthiemonge | - bumping and capping the requirements: https://review.opendev.org/c/openstack/octavia/+/905662/1/test-requirements.txt | 13:42 |
| gthiemonge | ? | 13:42 |
| tweining | I'd prefer the latter | 13:43 |
| tweining | although... you said stable branches. in that case I guess it is probably fine either way. | 13:48 |
| opendevreview | Takashi Kajinami proposed openstack/python-octaviaclient master: Bump hacking https://review.opendev.org/c/openstack/python-octaviaclient/+/905659 | 16:17 |
| johnsom | Yeah, I would prefer to move the version forward if it passes | 16:23 |
| opendevreview | Takashi Kajinami proposed openstack/octavia master: Cap hacking https://review.opendev.org/c/openstack/octavia/+/905660 | 16:57 |
| opendevreview | Takashi Kajinami proposed openstack/octavia stable/yoga: Stable-only: Pin hacking to < 6.1.0 https://review.opendev.org/c/openstack/octavia/+/905719 | 17:15 |
| opendevreview | Takashi Kajinami proposed openstack/octavia stable/xena: Stable-only: Pin hacking to < 6.1.0 https://review.opendev.org/c/openstack/octavia/+/905720 | 17:17 |
| spatel | johnsom did you see this error - https://ibb.co/zR4TscQ | 19:49 |
| spatel | openstack role add --project myproject --user-domain Default --user myuser1 load-balancer_member | 19:50 |
| spatel | I did run above command to add into role but still getting error | 19:50 |
| johnsom | Well, there could be a few reasons. One was openstacksdk broke with keystone. How new of a deployment is this? | 19:50 |
| spatel | This is new deployment but when I spin up k8s cluster it creates LB etc... | 19:51 |
| spatel | but when enduser try to create LB then throwing error | 19:51 |
| johnsom | See if these errors match yours: https://bugs.launchpad.net/octavia/+bug/2046711 | 19:52 |
| johnsom | Greg has a fix for that issue here: https://review.opendev.org/c/openstack/octavia-dashboard/+/903856 | 19:52 |
| johnsom | But that may not be your issue. | 19:52 |
| johnsom | That role looks fine to me, but there have been a bunch of RBAC changes in Openstack over the last two years that are... funky | 19:53 |
| spatel | reading... I have zed release which is working fine.. This is 2023.1 release causing issue.. | 19:55 |
| johnsom | Yeah, this breakage in openstacksdk is fairly new | 19:55 |
| spatel | hmm | 19:58 |
| johnsom | What is the error in horizon_error.log? That should point us in the right direction | 19:58 |
| spatel | let me check logs | 19:58 |
| spatel | hold on | 19:59 |
| spatel | [Tue Jan 16 20:02:46.626689 2024] [wsgi:error] [pid 24:tid 140187792516672] [remote 192.168.8.14:59002] Forbidden: /api/octavia-barbican/secrets/ | 20:02 |
| spatel | this is keep pop'ing | 20:03 |
| spatel | look like its barbican related issue.. | 20:03 |
| johnsom | Oh, ok, yeah, that is another issue. Give a minute | 20:03 |
| johnsom | https://review.opendev.org/c/openstack/octavia-dashboard/+/887027 | 20:04 |
| johnsom | Also, I think users need a "consumer" role in barbican to use it. Hmm, I wonder if that is a bug in this patch | 20:05 |
| spatel | so this is horizon patch | 20:05 |
| spatel | ? | 20:07 |
| johnsom | Hmm, no doesn't appear to be the issue as that patch is only in 2023.2 | 20:08 |
| johnsom | Hmmm, looking at that error again, that is odd, it's like horizon isn't allowing access to the internal secrets endpoint (the barbican library) | 20:11 |
| spatel | hmmm | 20:12 |
| spatel | I remember now.. Octavia was working fine.. until I install barbican | 20:13 |
| spatel | So definitely barbican broke something | 20:13 |
| johnsom | Yeah, that changes how horizon works. You may need to add a role to the user for barbican. Try "consumer" | 20:13 |
| spatel | add role consumer? | 20:14 |
| johnsom | openstack role add --project myproject --user-domain Default --user myuser1 consumer | 20:14 |
| johnsom | I think that is what they called it. | 20:14 |
| spatel | let me check first if role already exist or not | 20:15 |
| johnsom | Oh, looking at barbican docs, it might be "creator" | 20:15 |
| spatel | No consumer role | 20:15 |
| spatel | hmmm | 20:16 |
| spatel | There is a role called creator | 20:16 |
| johnsom | Yeah, try that | 20:16 |
| spatel | ok | 20:17 |
| spatel | it works :) | 20:18 |
| spatel | no error | 20:18 |
| johnsom | Yeah, so some oddness with how barbican sets up it's roles. | 20:19 |
| spatel | :( indeed | 20:20 |
| spatel | every rev come with some extra pieces in puzzle | 20:20 |
| johnsom | Oh I know.... It's a lot of work to track down who broke what in which repo.... lol | 20:24 |
| spatel | again, thank you for the help :) | 20:25 |
| johnsom | Sure, NP | 20:25 |
| spatel | johnsom is it possible to created single role and add creator and load-balancer_member role in it? | 20:40 |
| johnsom | Yes | 20:41 |
| spatel | so in single command we can assign both role | 20:41 |
| johnsom | I typically setup a single "user" role that has everything in it that all users need | 20:41 |
| johnsom | spatel https://docs.openstack.org/octavia/latest/configuration/policy.html#keystone-group-roles | 20:42 |
| johnsom | I wrote it up in our docs | 20:42 |
| spatel | This is awesome | 20:44 |
| spatel | so create group and assign roles to that group right? | 20:45 |
| johnsom | yep, if a group doesn't already exist | 20:45 |
| spatel | later add user to that group :) | 20:45 |
| spatel | I will give it a try | 20:45 |
| opendevreview | Mohammed Naser proposed openstack/octavia master: Revert "Fix remaining usage of [neutron] endpoint_type" https://review.opendev.org/c/openstack/octavia/+/905724 | 21:23 |
| opendevreview | Mohammed Naser proposed openstack/octavia master: fix: use oslo_conf when getting neutron client https://review.opendev.org/c/openstack/octavia/+/905794 | 22:13 |
| mnaser | https://bugs.launchpad.net/octavia/+bug/2049551 | 22:21 |
| mnaser | anyone ran into something like this? | 22:21 |
| mnaser | i've been stuck for a few hours trying to unbork this. | 22:21 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!