Tuesday, 2022-04-05

*** rlandy is now known as rlandy|out00:50
opendevreviewIan Wienand proposed opendev/gerrit master: Retire repo  https://review.opendev.org/c/opendev/gerrit/+/83394303:20
ianwfor reference, this is the list of branches to remove from opendev/gerrit, as discussed in the last meeting https://paste.opendev.org/show/b9Ic1Cpj55IM6arWe2ek/03:31
opendevreviewMerged opendev/gerrit master: Retire repo  https://review.opendev.org/c/opendev/gerrit/+/83394304:01
ianwok, repo is retired now.  we'll keep an eye if we need to purge those changes from master; if you end up at opendev.org when looking at gerrit things that would be a good data point04:26
opendevreviewDr. Jens Harbott proposed openstack/diskimage-builder master: Test building jammy  https://review.opendev.org/c/openstack/diskimage-builder/+/83622805:55
fricklerinfra-root: seems the twitter part of status bot hasn't been working for almost a year, is that something we still care about? otherwise we might want to mark it as abandoned and remove references from our docs https://twitter.com/openstackinfra06:00
ianwhuh, i did think that it was working post containerisation, but admit i don't keep close tabs06:10
ianw  File "/usr/local/lib/python3.8/site-packages/twitter/api.py", line 4907, in _ParseAndCheckTwitter06:17
ianw    raise TwitterError({'Unknown error': '{0}'.format(json_data)})06:17
ianwit probably died post python3 upgrade06:17
ianwlooks like it's this bug06:22
fricklerthere is a bug in gerrit regarding to attention sets: I'm marked as needing to look at https://review.opendev.org/c/zuul/nodepool/+/836263 since I uploaded a previous PS, but since I'm not the current uploader, I cannot remove myself any longer, so this patch is stuck in my "Your Turn" list06:22
fricklero.k., manually adding myself to the CC: list seems to be a workaround06:24
fricklerianw: so it looks like there isn't a solution yet?06:25
ianwlooks like all attention has moved to tweepy06:28
fricklerseems etherpad just went down? checking the server now06:36
fricklerlooks like log rotation hickup, came back without me actually doing anything except checking the docker log06:50
*** icey_ is now known as icey06:59
ttxLooks like ethercalc is down07:22
ttxfrickler, ianw: that prevents access to the PTG host keys07:22
ttx(https://ethercalc.openstack.org/ returns 503)07:23
*** jpena|off is now known as jpena07:37
ianwhrm, let me see07:37
ianwttx: ok now?07:40
ianwwhen i logged in there was no ethercalc node service running07:41
ianwApr 05 06:30:37 ethercalc02 bash[3586]: /opt/ethercalc/node_modules/redis/index.js:60207:44
ianwApr 05 06:30:37 ethercalc02 bash[3586]:                 throw err;07:44
ianwthat's where it died07:44
ttxfamous last words08:05
fricklerttx: should be stable again now, please let us know if you see further issues08:20
*** ysandeep is now known as ysandeep|lunch08:28
ttxfrickler: thanks!08:36
*** ysandeep|lunch is now known as ysandeep09:00
*** pojadhav is now known as pojadhav|afk09:50
*** pojadhav|afk is now known as pojadhav10:12
*** rlandy|out is now known as rlandy10:27
fungipuppet also would have restarted it on the next periodic deploy run, since the service resource has an ensure=>running11:11
fungiunfortunately it crashes a lot11:11
opendevreviewAndy Ladjadj proposed zuul/zuul-jobs master: Update upload-logs azure role to support endpoint override  https://review.opendev.org/c/zuul/zuul-jobs/+/83664012:04
opendevreviewAndy Ladjadj proposed zuul/zuul-jobs master: Update upload-logs azure role to support endpoint override  https://review.opendev.org/c/zuul/zuul-jobs/+/83664012:06
opendevreviewAndy Ladjadj proposed zuul/zuul-jobs master: Update upload-logs azure role to support endpoint override  https://review.opendev.org/c/zuul/zuul-jobs/+/83664012:27
*** artom__ is now known as artom13:28
zigoThere's no release notes for Yoga online, at least for puppet-keystone: https://docs.openstack.org/releasenotes/puppet-keystone/unreleased.html13:29
zigoHow can this be fixed?13:29
fungiif a change merges for puppet-keystone, the release notes job should run again and publish content there. did the 20.3.0 tag not upload any release notes? i'll double-check whether that job ran13:34
fungilooks like it ran successfully: https://zuul.opendev.org/t/openstack/build/ab6f733179c34a6699f3d5e5fa78e17c13:36
fungii think it may be a problem with the design of the job. there was a yoga releasenotes page but the publication of the 20.3.0 tag's releasenotes deleted it: https://zuul.opendev.org/t/openstack/build/ab6f733179c34a6699f3d5e5fa78e17c/console#4/0/7/localhost13:38
fungihttps://review.opendev.org/833960 was the release request which triggered that13:39
fungilooks like the stable/yoga branch was created at the same time as the tag, so i wonder if there's a race with the new branch if the tag is pushed at the same moment it's created13:41
fungizuul did provide a stable/yoga branch for the puppet-keystone repo though: https://zuul.opendev.org/t/openstack/build/ab6f733179c34a6699f3d5e5fa78e17c/console#0/4/12/ubuntu-focal13:43
fungilooks like the yoga releasenotes should automatically get recreated once the next change merges for that repo, but it would be good to figure out why reno didn't build yoga release notes in that run13:46
Clark[m]fungi: re tox and use develop the package is installed by default without use develop. https://tox.wiki/en/latest/config.html#conf-skip_install you need to set skip_install to skip that14:15
fungiClark[m]: it really isn't though. if skipsdist is true and usedevelop is false or unset, the resulting env doesn't have the project in it unless you explicitly add it to deps (i just tested again with latest tox to confirm my recollections)14:28
fungii have a project where i'm obsessive about not installing it into any tox testenvs which don't absolutely require it to be installed (in order to also avoid dragging in its requirements), so for the testenvs where the project does need to be installed i achieve that by explicitly adding {toxinidir} to the deps list in the testenv14:30
fungithe rest don't end up with the projects or its requirements in the resulting venvs14:30
Clark[m]Oh skipsdist is like skip_install then?14:40
Clark[m]I thought it was just to pip install . Rather than the tarball14:41
Clark[m]I think that is a behavior change from when I last looked at these settings (years ago I'm sure)14:42
fungiskipsdist avoids explicitly building an sdist of the project, which would otherwise have a side effect of installing the project14:43
fungiin the example project, i have a separate testenv:dist which i use to test that sdist creation works (as well as wheel building, and performs some cursory analysis of the resulting artifacts)14:45
Clark[m]Right but skip_install is a separate setting and I'm pretty sure that skipsdist only controlled the building of the sdist in the past. If you set it in the past it would still install the project just without a separate sdist step14:45
Clark[m]Which would do whatever pip does to install the project rather than a pip install of the resulting sdist iirc14:46
fungii actually don't recall that being the case. usedevelop also installs the project, which probably led to some confusion. if skipsdist was true *and* usedevelop was false (or unset) then you didn't get an automatic install of the project, but if at least one of those was not the case then it ended up getting installed14:46
fungii ran into it back when we were first working on bindep adoption, because i wanted it to be possible to `tox -e bindep` in a nova checkout without nova or its requirements getting installed. unsetting usedevelop and making skipsdist true accomplished that14:48
fungiunfortunately, some time later, nova dropped the skipsdist setting from their tox.ini, so now it's no longer working as intended14:50
clarkbya testing this it works as you say. Its weird to hvae the separate setting which is cimpletely ignored14:50
clarkbfungi: is it worth noting on the thread that the behavior you describe depends on skipsdist?14:51
fungithey also edited my comment in their tox.ini which said tox.skipsdist = True was necessary :/14:51
clarkbI guess the assumption is that openstack sets that fairly universally?14:51
fungihttps://review.openstack.org/622972 (erroneously) removed skipsdist, and then https://review.opendev.org/684775 later cleaned up the comment which said it was needed14:52
fungii think there's a lot of misconceptions about how those settings work, leading to projects randomly adding/removing them in hopes the black magic will work14:53
fungii wouldn't assume anything at this point14:53
*** pojadhav is now known as pojadhav|afk14:56
clarkbfungi: might be worth a followup to the thread then? Just to make it clear that usedevelop=true is only responsible for that behavior when skipsdist is also true. Otherwise default tox behavior can exhibit this as well14:56
fungiyep. i'll try to add something once i get a break14:57
clarkbthanks. Ya I'm about to jump on the call I expect you are on. I just have to figure out how to connect from my phone14:58
fungii'm joining the qa ptg room for the fips testing discussion, yes14:58
slittleplease add me as first member of group starlingx-app-istio-core15:03
slittleI can add the rest after that15:03
fungislittle: will do, just a moment15:04
fungislittle: done15:06
fungiclarkb: are you going to be in the qa room for the openstack-health topic? it will probably be when i need to be talking about xstatic packages in the horizon room15:39
clarkbya I can be15:39
clarkbit was on my rough list of things to be around for15:40
funginot sure whether they need someone from opendev/tact-sig in there but figured it might be in your wheelhouse15:40
fungicool, thanks!15:40
*** marios is now known as marios|out15:47
opendevreviewMarcin Juszkiewicz proposed opendev/system-config master: mirror-update: added EPEL 9  https://review.opendev.org/c/opendev/system-config/+/83667015:48
*** pojadhav|afk is now known as pojadhav15:58
*** ysandeep is now known as ysandeep|out16:03
opendevreviewMarcin Juszkiewicz proposed opendev/system-config master: mirror-update: added EPEL 9  https://review.opendev.org/c/opendev/system-config/+/83667016:12
*** jpena is now known as jpena|off16:31
clarkbnoting this here but openstack jobs have started to see failures due to the pypi served us a stale index file for $package problem18:15
clarkbit seems to be affecting inmotion iad3, rax iad, and iweb mtl01 (all east coast north america ish) and a number of packages18:16
clarkbat this point based on debugging I've done I don't think any of our services are the source of the problem and instead this is what we've debugged in the past where pypi's CDN will fallback to grabbing info from a backup source that is nowhere near up to date which means newre packages are not present in indexes18:17
clarkbalso TIL openstack uses a package called "itsdangerous"18:22
fungiahh, data integrity18:33
*** rlandy is now known as rlandy|mtg18:33
*** rlandy|mtg is now known as rlandy19:02
opendevreviewMerged opendev/system-config master: Finish removing git.openstack.org references  https://review.opendev.org/c/opendev/system-config/+/65404019:50
clarkbfungi: are you aware of wednesday sessions we should be in? I'm planning to do the thursday and friday TC stuff I think20:01
clarkbor at least as much as I can manage with my morning time20:01
fungiclarkb: none i'm aware of, the only thing i have for tomorrow is barbican consumers api discussion since it's related to/blocking the image encryption effort20:06
TheSovtrying to use gerrit from my desktop and keeping get public key errors, any idea why that is, i made sure the key matched several times20:08
clarkbTheSov: chances are you are on a system that doesn't allow sha1 + rsa keys (fedora, opensuse tumbleweed, or anything with a very new openssh)20:09
TheSovusing ubuntu wsl20:09
clarkbWe think this will be fixed in gerrit 3.6 whenever it releases. Until then you can tell openssh to allow sha1 keys or use a non rsa key. We've been suggesting a non rsa key20:09
clarkbhttps://www.openssh.com/txt/release-8.8 documents allowing sha1 with rsa under Potentially-incompatible changes20:10
TheSovthanks for the direction!20:10
TheSovill try it20:10
*** rlandy is now known as rlandy|mtg20:26
*** dviroel is now known as dviroel|out20:36
opendevreviewGhanshyam proposed openstack/project-config master: Remove tempest-lib from infra  https://review.opendev.org/c/openstack/project-config/+/83670322:45
clarkbianw: did you see the change to add epel mirroring for centos 9? https://review.opendev.org/c/opendev/system-config/+/836670 I tried to give it proper review both for afs room and appropritae content. It looks good to me22:49
*** rlandy|mtg is now known as rlandy23:16
ianwclarkb: thanks, will look23:21
fungiianw: also in case you missed scrollback in #openstack-infra, seems there are problems with the state of the f35 mirror. the refreshes have shown a bunch of churn, so hopefully it'll settle down within the coming hours23:26
fungimight be worth keeping an eye on though23:26
ianwyeah i noticed that, and i agree with your analysis of a partial update23:28
ianwas you say, i'll check in a bit and see if it looks better23:28
opendevreviewGhanshyam proposed openstack/project-config master: Retire openstack-health project: end project gating  https://review.opendev.org/c/openstack/project-config/+/83670723:43
clarkbgmann: you'll need a change in health to update the .zuul.yaml23:44
clarkbI guess we have to do noop jobs in project-config so they apply when we delete the content23:44
gmannclarkb: yeah, I am on the next step to remove the content in repo which need noop job ^^23:45
clarkbyup I approved it I think single core on that one is fine23:46
gmann+1. thanks. 23:48
*** rlandy is now known as rlandy|out23:55
clarkbheh all the extra people in my house will be gone tomorrow but its the one day this week where I don't need todial into ptg calls23:58
clarkbdidn't plan that very well23:58

Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!