Friday, 2022-03-25

opendevreviewMohammed Naser proposed zuul/zuul-jobs master: ensure-kubernetes: fix missing 02-crio.conf
opendevreviewMohammed Naser proposed zuul/zuul-jobs master: run-buildset-registry: Drop extra install packages task
fungilooks like we're just waiting for ze01 at this point01:47
*** ysandeep|out is now known as ysandeep04:58
*** chandankumar is now known as chkumar|rover05:52
*** ysandeep is now known as ysandeep|afk06:13
*** ysandeep|afk is now known as ysandeep07:01
*** ysandeep is now known as ysandeep|lunch08:04
*** marios is now known as marios|out08:07
*** arxcruz|ruck is now known as arxcruz08:10
*** jpena|off is now known as jpena08:11
*** marios|out is now known as marios08:18
*** marios is now known as marios|out08:27
*** mgoddard- is now known as mgoddard09:11
*** ysandeep|lunch is now known as ysandeep09:32
opendevreviewdaniel.pawlik proposed opendev/system-config master: Add Opensearch Dashboard information
ysandeepfyi.. latest setuptools broke tripleo-ci check/gate jobs, similiar issue to what's reported here: , We are trying to workaround those issue as per github issue.. But just incase other projects hit this as well.09:46
opendevreviewdaniel.pawlik proposed opendev/system-config master: Add Opensearch Dashboard information
*** rlandy|PTO is now known as rlandy10:28
priteauHello. Is anyone aware of issues with py39 tests today? blazar py39 tests are timing out in CI, they run fine locally10:45
priteau proposed openstack/diskimage-builder master: Add interpolation note for dynamic-login password
fricklerpriteau: might be related to the setuptools issue tripleo is seeing? everyone else seems to be fine
priteauI am not sure, they have a specific error message11:02
priteauWe haven't run jobs on master recently, so it broke sometime between March 10 and yesterday11:05
priteauUpper constraints shouldn't have changed much since then. Very odd.11:07
fungii doubt a tox-based job would be picking up yesterday's setuptools yet. usually that needs new virtualenv first11:42
fungiyeah, looks like that job used 60.10.0 not 61.0.011:44
fungialso that first timed_out result was before the new setuptools release happened anyway11:47
*** chkumar|rover is now known as raukadah12:01
*** rlandy is now known as rlandy|rover12:30
rlandy|roverfungi: hello ... can you force merge a patch for us to clear the current tripleo gate blocker?
rlandy|roverwe have ptl approval13:58
rlandy|roverysandeep: jm1: ^^13:59
rlandy|roverjpodivin: hello14:00
rlandy|roverjust asked above for force merge14:00
fungirlandy|rover: yep, working on it now14:03
rlandy|roverfungi: great - thanks14:03
fungirlandy|rover: do you just need it moved to the front of the gate, or more?14:03
rlandy|roverfungi: it will fail in gate14:04
rlandy|roverwe have a bunch more patches to merge14:04
rlandy|roverfungi: we also need the one in gate right now 835178,514:04
rlandy|roverwe need that one to finish14:04
rlandy|roverit should pass14:05
fungirlandy|rover: they can't be squashed together? they're for different repos i guess?14:05
rlandy|rovercollect logs and tripleo-ci14:05
fungigot it.14:05
rlandy|roverthen we will have to address quickstart and a few otheres14:05
rlandy|roverlovely morning we are having :(14:05
rlandy|rover835178,5 will also need to force merge if it fails - but it shouldn't14:07
fungi#status log Bypassed Zuul to force the merging of 835193,10 due to circular dependency on changes for related repositories addressing regressions in Setuptools 61.0.014:10
opendevstatusfungi: finished logging14:10
*** artom__ is now known as artom14:14
rlandy|roverfungi: look sto have merged - thank you14:15
fungiyep, see above14:15
rlandy|roverfungi: awesome  - much appreciated!!14:15
fungiyou're welcome!14:15
fungigood luck untangling the rest14:15
*** diablo_rojo_phone is now known as Guest31814:42
*** ysandeep is now known as ysandeep|dinner14:46
corvusfungi: i'm going to be afk for a chunk of the day... i could do the rest of the zuul restart now (schedulers/web), or do it after i return... which gives you the warm fuzzies?14:52
fungicorvus: now is fine!14:53
corvusmaking it so14:53
fungii saw the mergers and executors finished but i didn't want to fire off the rest in case you had something specific you wanted to check14:53
corvusnah, more concerned with the unexpected (ie, a usual deployment)14:54
fungii actually fell asleep before the first batch of executors finished last night anyway14:54
Clark[m]I'll be around today. Just trying to get started14:56
corvus1/2 done15:03
corvus#status log rolling restart of all of zuul on at least 5.2.015:06
opendevstatuscorvus: finished logging15:06
corvusfrickler: is still useful?15:07
clarkbfungi: for the setuptools issues sounds like the impacted groups are aware and addressing it. Otherwis ewe haven't seen problems?15:10
fricklercorvus: I must admit that I don't even remember anymore what I was testing there, will clean up, thx for spotting15:13
corvusfrickler: np, i can delete it if you want15:13
corvusis one click in web ui :)15:14
fricklerah, that was pluggy vs. py2. deleted.15:16
corvusthe restart is completely complete now :)15:24
fricklerbtw., regarding zuul auth, do we have docs how the auth_token in zuul02:/root/.config/zuul/client.conf was created? better be prepared when it expires in 9 years15:24
corvusjust the upstream:
fungiclarkb: yeah, so far only tripleo is confirmed impacted, i think because they install setuptools directly and aren't pinning it. devstack still may be pinning setuptools, not sure. tox isn't using the new setuptools yet because it depends on virtualenv's vendored copy15:32
*** ysandeep|dinner is now known as ysandeep15:38
*** ysandeep is now known as ysandeep|out15:44
clarkb#status log Deleted 9 pastes from paste.o.o at the request of the users.16:04
opendevstatusclarkb: finished logging16:04
fricklercorvus: is that the same type of token as described in or are those two different things?16:11
dtantsurhey folks, rings any bells? mkswap: error: /root/swapfile is mounted; will not make swapspace16:27
clarkbno that is new to me16:28
clarkbthe issues we've had in the past are using a sparse file which latest linux kernels (f35) don't support anymore so we switched to dd'ing zeros16:28
dtantsur has more than a dozen RETRY_LIMITs16:29
clarkbdtantsur: I suspect that something else made a swapfile16:29
clarkband then when that got there it failed16:29
dtantsur and
dtantsuryou're right16:29
clarkbyes that is what happened16:30
dtantsurI don't get it... is run, then
dtantsurit's not grenade, why are two versions executed?16:31
clarkbyou probably have an explicit branch matcher16:33
clarkband now that yoga is branched it is matching multiple branches16:33
clarkbexplicit branch matchers should almost never be used on branched repos but people do add them and they create these problems when branches are made. The trick now is to hunt it down16:34
dtantsurwe do have to use them sometimes, but not in this case? at least I cannot immediately spot anything that affects all jobs16:34
dtantsur(and I'm not even sure I understand "branch matchers" correctly)16:35
clarkbdtantsur: a specification on a job that indicates which branch the job should run against16:35
clarkbif your repo is branched you should rely on the implicit matching that says the definition on this branch runs against hte branch and not override that16:35
clarkblet me take a look at the job definitions16:36
dtantsurclarkb: if I understand it right, we do it for bugfix branches, but not for master16:36
clarkbdtantsur: that is the problem. I feel like I had to fix this for devstack when xena was branched too :/16:40
clarkblet me cross check against that and push a change up16:40
dtantsurahhh. wow. I haven't seen this syntax before.16:40
dtantsurI thought you were talking about override-checkout or something like that16:41
clarkbdtantsur: not override-checkout but branches:16:41
dtantsurI see. Never used that, I think.16:41
clarkbbut ya there is this other thing that devstack does and I thought we had made sure that the branching process included addressing this for next time16:41
clarkbdtantsur:remote: Remove the zuul pragma to match master and feature/r116:44
dtantsurdo we need the same patch for stable/yoga?16:53
clarkbdtantsur: so my change is wrong because it went to mater because we haven't updated .gitreview yet. There is an approved change for this already though. I'll see if that needs help landing16:53
clarkbI'm going to abandon the chnage I just pushed16:53
clarkblooks like it is in the gate so just a matter of waiting now. Depends-On should work too
fungishould we promote it, or is there not much backlog?16:55
clarkbnot much backlog. Its near the top and promoting would just prolong right now I think16:56
fungiyeah, if it was waaay down with a bunch of changes failing ahead on the same problem then it would be a different situation16:57
*** jpena is now known as jpena|off17:06
clarkblooks like nodepool (and I assume zuul soon too) are looking at doing python 3.10 testing but I don't think our opendev base image has a 3.10 variant yet. We might want to update as much as we can to python3.8 or 3.9 then drop 3.7 and add 3.10 images?17:49
opendevreviewClark Boylan proposed opendev/system-config master: Update gitea to v1.16.5
clarkbif testing says ^ that is happy it should be a fairly safe update. But I also don't think it is super urgent18:56
clarkbthe screenshots from 835307's deployment lgtm20:12
clarkbI'm reading up on the setuptools discovery stuff and I expect many things will be fine as they follow the flat-layout/adhoc structure21:02
clarkbWhere we might have trouble is if we put python scripts in playbooks/ or roles/ dirs21:03
clarkboverall though I don't expect widespread fallout21:03
clarkbdoc/ tools/ tests/ are all excluded by default. However that also means they will be excluded from sdists/wheels21:04
clarkbwhcih we may want to avoid?21:04
clarkbI think system-config might break. But where it breaks should be limited: doc builds?21:08
clarkbsomething like `find ./ -type f -name '*.py' | grep -v ./doc/ | grep -v ./tools/ | grep -v './test[^/]*/'` where you also ignore $packagename dir to find problems21:10
clarkbI'm not able to make system-config tox -e docs fail if I upgrade the setuptools install in the .tox/docs dir and rerun21:13
clarkbmaybe we are ok there then21:13
clarkbhrm nevermind if I manually construct a new virtualenv with up to date setuptools and try to pip install setuptools from there it fails21:21
clarkb*try tp pip install system-config form there21:21
clarkbfiddling with the setup.cfg to override auto discovery I'm not quite sure what the correct thing to do here is. The docs have a good example but for a completely different setup than ours21:32
clarkbI'm also not entirely sure the fix tripleo used is correct? setting py_modules to [] should result in emptyish packages?21:45
clarkbon the system-config side of things setting 'packages = ' seems to produce wheels that look like the wheels produced by old setuptools (and are very empty)21:46
clarkbI think beacuse pbr is looking for in top level dir to find the source dir?21:46
fungii think sticking with pbr means we won't need to care about setuptools discovery functionality21:47
fungibut system-config could be an outlier21:47
fungipbr uses git's tracked files to build the itself, right?21:48
clarkbya it should and checking tripleo-quickstart using old setuptools and new setuptools against their current fixed code produces wheels iwth 300 files each and almost the same total size21:52
clarkbI suspect that in this case we can lie about the py_modules and/or packages lis tand pbr will figure it out 21:53
clarkbIt might be worth understanding that better and leaving comments in the setup.cfg/ filesso there isn't confusion over this21:54

Generated by 2.17.3 by Marius Gedminas - find it at!