Thursday, 2022-01-27

*** ysandeep|out is now known as ysandeep06:05
*** dmsimard6 is now known as dmsimard06:38
*** amoralej|off is now known as amoralej07:56
ykarelopendevreview bot died?08:12
*** ysandeep is now known as ysandeep|lunch08:30
*** jpena|off is now known as jpena08:37
fricklerykarel: seems it lost its IRC connection without actually noticing. restarted it now08:37
frickler#status log restarted gerritbot which had gone missing at 04:56:4008:39
opendevstatusfrickler: finished logging08:39
gthiemongeHi Folks, I would like to investigate CI issues, is it possible to hold a node so I could ssh to it?08:41
fricklergthiemonge: sure, job name and patch ID, please ;)08:43
gthiemongefrickler: patch id: 816370 jobname: octavia-v2-dsvm-scenario-centos-9-stream08:48
gthiemongeoops, it failed for an unrelated reason :/08:48
fricklergthiemonge: so - no hold needed for now?08:52
gthiemongefrickler: yeah, probably a mirror issue with Centos, I'll ping you when this step passes, thanks!08:55
ykarelThanks frickler 08:57
fricklerinfra-root: is weird to me, recheck triggers gating instead of running checks again first, not sure if that's intentional or some kind of bug.09:01
*** ysandeep|lunch is now known as ysandeep09:17
hrwin next week Kolla project switches to using Yoga packages. adds mirroring of canonical UCA repo for Yoga.10:16
opendevreviewAlfredo Moralejo proposed zuul/zuul-jobs master: Add CentOS Stream 9 to configure-mirrors role
opendevreviewAlfredo Moralejo proposed zuul/zuul-jobs master: Add CentOS Stream 9 to configure-mirrors role
*** sboyron_ is now known as sboyron10:44
opendevreviewAlfredo Moralejo proposed zuul/zuul-jobs master: Add CentOS Stream 9 to configure-mirrors role
*** rlandy|out is now known as rlandy|ruck11:01
*** bhagyashris_ is now known as bhagyashris|ruck11:05
gthiemongeFolks, could you hold the job octavia-v2-dsvm-scenario-centos-9-stream for change id 816370 for me?11:17
*** dviroel|afk is now known as dviroel11:20
fricklerinfra-root: ^^ I now get: sudo docker exec -it zuul-scheduler_scheduler_1 zuul autohold-list --tenant openstack11:21
fricklerUnable to find a way to connect to Zuul, add a "web" section to your configuration file11:21
fricklerwhat's the new way of doing that? if we have already documented that somewhere, I missed it11:21
opendevreviewAlfredo Moralejo proposed zuul/zuul-jobs master: Add CentOS Stream 9 to configure-mirrors role
*** sshnaidm|afk is now known as sshnaidm11:35
fricklerah, it's simply "sudo zuul-client something ..."12:19
opendevreviewAlfredo Moralejo proposed zuul/zuul-jobs master: DNM test only
fricklerthe quoting is broken the way it is being passed into that docker command, though12:24
fricklergthiemonge: your hold is in place now, it should catch the job that is currently running if it fails. what's your ssh pubkey?12:25
amoralejhi, it seems tox-py27 is broken in zuul-jobs
amoralejERROR: Package 'oauthlib' requires a different Python: 2.7.18 not in '>=3.6'12:28
amoralejis this known issue?12:28
opendevreviewDr. Jens Harbott proposed opendev/system-config master: Fix quoting in zuul-client docker wrapper
fricklergthiemonge: root@ , let us know once you are done debugging12:44
gthiemongefrickler: thanks12:44
hrwamoralej: update pip to 20.* and retry12:45
hrwamoralej: if this is c7 then pip 8 is used which tries to download whatever and then complain it fails12:46
amoralejno, it seems that runs on ubuntu-focal12:46
amoralejit was working until yesterday
amoraleji though we wouldn't be testing py27 at this point12:47
hrwamoralej: oauthlib 1.3.1 is fetched in newer jobs. and it dropped py2 code12:49
hrwwheel file is py2.py3 but there is no py2 inside12:49
amoralejbut it seems to be in the mirrors since december12:49
hrwupper-constraints to the rescue ;(12:49
amoralejshould i pin for python2.7?12:49
amoralejmmm do we use u-c there, let me check12:49
amoralejactually seems related to the version of pip or tox12:50
amoralejin my fedora it pulls 3.1.012:50
hrw3.1.0 I meant and 3.1.112:50
hrw3.1.0 is last really py2.py312:50
amoralejyep 3.1.0 is fine12:51
amoralejbut it's pulling 3.1.1 since today12:51
hrwI treat py27 jobs as ugly legacy where pinning is allowed as they will be dropped 'real soon now'12:51
hrwtook me 3 days to fix stable/train in kolla - mostly 2.7 fixes and c7/c8 handling12:52
opendevreviewAlfredo Moralejo proposed zuul/zuul-jobs master: Add CentOS Stream 9 to configure-mirrors role
amoralejthanks hrw !12:55
amoralejlet's see how it goes12:55
opendevreviewAlfredo Moralejo proposed zuul/zuul-jobs master: Add CentOS Stream 9 to configure-mirrors role
hrwamoralej: I wonder why tox-py27 is called in cs9 jobs12:57
amoralejhrw, is called in all changes in zuul-jobs12:57
hrwyou do yoga time change when py27 is history and should be forgotten12:57
hrwyou need to add u-c change to it13:00
hrwor sth like that13:00
* hrw out13:00
amoraleji did it in
opendevreviewAlfredo Moralejo proposed zuul/zuul-jobs master: pin oauthlib version for python2.7
gthiemongefrickler: I'm done with, thank you, it helps me a lot!13:13
*** amoralej is now known as amoralej|lunch13:23
fricklergthiemonge: welcome, node released13:27
opendevreviewAlfredo Moralejo proposed zuul/zuul-jobs master: pin oauthlib version for python2.7
opendevreviewDr. Jens Harbott proposed zuul/zuul-jobs master: pin oauthlib version for python2.7
frickleramoralej|lunch: oops, my edit clashed with yours, I'll restore13:34
opendevreviewDr. Jens Harbott proposed zuul/zuul-jobs master: pin oauthlib version for python2.7
*** ysandeep is now known as ysandeep|brb13:35
opendevreviewDr. Jens Harbott proposed openstack/project-config master: Drop gerritbot from #osism channel
fungifrickler: yeah, the /usr/local/bin/zuul-client on the schedulers is a wrapper around docker run already. and i agree, it would be nice if we could get multi-word parameter quoting to work reliably. i managed with escaping my quotes last time i needed to do it13:49
fricklerfungi: see 826653, I tested that successfully on zuul0213:50
fungioh, awesome--thanks!13:51
fricklerI always forget which way around that works, so I added a link to the bash docs where I found it again13:51
fungiyep, perfect. i'm the same way with $@ vs $*13:52
*** ysandeep|brb is now known as ysandeep14:03
*** amoralej|lunch is now known as amoralej14:23
opendevreviewAlfredo Moralejo proposed zuul/zuul-jobs master: Add CentOS Stream 9 to configure-mirrors role
*** artom__ is now known as artom14:50
fungiopenstack tc meeting is starting now, and the agenda includes an entry to discuss the possibility of renaming the primary branches for all projects, which is something we've discussed the potential complexities of in the past. i'll do my best to reiterate our current situation and what we know so far15:00
*** Guest292 is now known as diablo_rojo_phone15:01
Clark[m]fungi: thanks. I won't be able to make the meeting. I do still think having a single repo do it first to sort out unexpected issues is a good first step. And better if we can find a new repo rather than a well entrenched one to be the first.15:15
fungiyes, for actual branch moves i've seen a number of projects use symbolic refs on their remotes, but i think that would be challenging in our case because of our replication model (those don't replicate)15:17
fungii think we'd have to independently create the refs on gerrit and every single gitea backend, and on github too if that mattered to people15:18
fungiClark[m]: summary is that lourot is going to help us test importing an unimportant new project with a default branch of main and see how things go, as a means of gathering more info and helping spot any unforeseen problems in our basic tooling15:44
fricklerdoes anyone have experience with ? just the other day I was once again lamenting about how much better I like the gerrit review process over github and that might be a low-key solution15:45
fungii've been aware of gerrithub for years, and i know they pioneered some of the high-availability multi-gerrit stuff, but i've never used their service15:45
corvusfrickler: it's run by folks who are deeply involved in the gerrit community15:51
*** dviroel is now known as dviroel|lunch16:13
opendevreviewMerged openstack/project-config master: Drop gerritbot from #osism channel
opendevreviewJeremy Stanley proposed opendev/system-config master: Drop wiki-dev03 from inventory
*** jpena is now known as jpena|off16:43
clarkbfungi: frickler if you have time to review and child I probably have time to monitor those today16:52
fricklerclarkb: +2 on the first but some comment on the second patch17:01
clarkbfrickler: perfect. I was far less confident in the second so questions are good (this is why I split them too)17:02
clarkbfrickler: aha I was trying to find where it came form on newer ubuntu and not finding it. I think you are right though we can just accept the overlap instead17:03
clarkbI'll update17:03
clarkbinfra-root I think I've been given the votes I need to merge the signed tag fix on gerrit 3.4, but there were questions around asking another reviewer to take a look first. I'm thinking I'll go ahead and land it early next week if the other reviewer doesn't get back to us and we can always work towards improvements in the future with them if necessary17:04
fricklerclarkb: "dpkg -S /etc/apt/apt.conf.d/10periodic" is what I did, though it might not be installed on our nodepool images17:14
opendevreviewClark Boylan proposed opendev/system-config master: Manage 10periodic and 20auto-upgrades together
clarkboh right dpkg will only query its local package list17:19
clarkbfrickler: ^ does that look better?17:19
*** ysandeep is now known as ysandeep|out17:21
hrwclarkb, fungi: can you review one? Adds mirroring of Yoga UCA for Focal17:26
*** dviroel|lunch is now known as dviroel17:26
clarkbhrw: re python2.7 earlier because zuul is a CI that people might want to use to test python2.7 despite its EOL we still need to test that zuul can run python2.7 jobs17:28
hrwclarkb: understood17:28
clarkbI agree that people should migrate away but even openstack is still running python2.7 tests and we have to support that17:29
clarkb(for old branches)17:29
hrwyeah, I spent 3 days in py27/train hell 17:29
hrwthis week17:29
hrwif py2.7 jobs fail then pip 20.* is kind of mandatory update for start17:30
*** amoralej is now known as amoralej|off17:30
opendevreviewMerged opendev/system-config master: Manage apt.conf.d/20auto-upgrades
fungiyeah, i'm still unclear on why pip 9 and later isn't sufficient, in cases where there's a python_requires excluding 2.7 for packages that don't support it17:34
fungipip 9 was when the python_requires support was integrated17:34
hrwit is all to '8-9 or 20' nowadays17:34
hrwand probably most of people building wheel files use latest possible17:34
clarkbhrm suddenlty I wonder if I should've confirmed yoga uca exists17:36
hrwit does17:36
clarkb yupconfirmed17:36
hrwthey created it during first weeks of Yoga cycle.17:36
hrwIn previous cycles I sent such patch much earlier but in Kolla we decided to switch to binary packages in R-8 week instead of as-soon-as-possible17:37
fungiclarkb: catching up on the repo-discuss thread about unresolved inline review comments, i have a feeling the majority of them in our projects are where i see you typoed something and make an inline comment pointing it out, then you push a new revision which fixes the thing i pointed out... gerrit has enough information to be able to tell whether the line i commented on changed since the17:47
fungipatchset where i made the comment, and could consider that resolution17:48
clarkbfungi: ya one person was suggesting that heuristic could be used17:48
fungior at least consider it an indication that the old comment should be hidden if it's not for either side of the current diff17:48
fungimaybe not explicitly resolved, just not shown (or maybe minimized)17:49
fungiif i comment on a line to point something out, and you push a new patchset without changing that line, then continuing to display the inline comment about it could make sense in that workflow17:50
clarkb++ we're transitioning the discussion to the issue tracker if you want to add any info17:50
fungii found the link, thanks17:50
clarkblooks good thanks and ya that might be a good simple rule they can apply18:06
*** sshnaidm is now known as sshnaidm|afk18:13
opendevreviewMerged opendev/system-config master: reprepro: mirror Ubuntu UCA Yoga
opendevreviewClark Boylan proposed opendev/system-config master: Rebuild Gerrit images particularly for 3.4
clarkbthat isn't particularly urgent but seems like a good idea18:29
clarkbwe'll need to do it again when the signed tag fix lands. As mentioned before I figured I'd give the additional reviewer until monday to weigh in if they like before proceeding on that one18:29
clarkbzuul is busy today18:38
fungiwish there was more movement on
clarkbfungi: maybe we should just clean up my fix (add docs updates etc) and push it up?18:39
fungimaybe if we pushed your "fix" up for review in their gerrit, they'd be more inclined to comment?18:39
fungihah, jinx18:39
fungithough it might make more sense to start with correcting the ${commit} macro behavior in gitweb.file and then pivot to a separate macro and doc updates if they decide that's a bad idea18:41
fungibut as you observed, the straightforward fix might have ripple effects in their gitiles integration18:41
clarkbya I'm pretty sure this will breka gitiles if we just change the type of commit18:44
clarkband who knows what other installations rely on this. I strongly suspect we need to add the new value18:44
clarkbLet me clean up that fix on 3.4 (it conflicts) and add a doc update and push that upstream18:44
clarkbalso I just realized I think we can depends on the chagnes when pushed to gerrit upstream18:45
clarkbthis should simplify testing and is worht a try18:45
fungioh! indeed, neat idea18:46
clarkbfungi: ok I've got that staged up locally. I wrote myself some TODOs for what needs doing before pushing upstream (they have a linter I need to run that takes me forever to figure out and I need to make sure this still builds on 3.4 etc)18:50
clarkbBut I've got a meeting in a few minutes so will probably happen after lunch18:51
fungioh, no rush, i'm just trying to decide whether/how to roll forward with gitiles disablement18:53
fungithanks for that!18:53
clarkbTheir linter changes the files in place and I don't like that bceause it makes it hard to understand what rules they want you to follow. But I guess most people don't mind because it makes it the way they need it to be18:53
clarkbbut also its a java tool from google that you have to figure out how to execute18:54
fungihopefully they have straightforward contributor docs18:54
fungii guess java linters aren't easily come by so they couldn't just reuse a more generic one18:55
clarkbya you basically download the jar from google and then run it directly18:58
clarkbor at least that is what the CI system did that first -1'd me so is what I've done since then18:58
clarkbthere may be a better method18:58
fricklerchecking the space needed by the yoga UCA (seems neglegible), I saw that ubuntu and centos both seem very close to their limits. can we clean something up there or do we have room to bump those limits?19:40
clarkbfrickler: we can delete centos8 in its entirety at the end of the month (that is when I announced we would do it)19:41
clarkbthat should free up quite a bit of space on the centos volume19:41
clarkbubuntu is more difficult and we might need to bump its quota19:42
clarkb shows we probably have some headroom to do that now19:42
clarkb(thats a lot more headroom than I expected too I guess the tumbleweed and other cleanups helped)19:42
clarkboh and I think fedora 34 is also in line to go away19:43
fricklercentos is at 348/350G and seems to be growing by about 1G/d. unrelated, project.opendev has a limit of 1T, which seems a bit excessive with 134M being used, was that meant to be 1G instead?19:47
fungii expect so, yes19:53
clarkbhuh I wonder why centos is growing like that20:07
clarkbI wonder if we are syncing a daily VM image or container image that we shouldn't be20:07
*** dviroel is now known as dviroel|out20:20
ianwfrickler: heh, had also fixed the quotes on that20:43
opendevreviewM Camp proposed openstack/project-config master: Update StarlingX docs promote job for R6 release
opendevreviewMerged opendev/system-config master: Run zuul project rename steps on a single scheduler
opendevreviewMerged opendev/system-config master: Fix quoting in zuul-client docker wrapper
opendevreviewMerged openstack/project-config master: Update StarlingX docs promote job for R6 release
*** hrww is now known as hrw23:08
*** rlandy|ruck is now known as rlandy|ruck|bbl23:26
opendevreviewMerged openstack/diskimage-builder master: Rename existing BLS entry with the new machine-id
clarkbfungi: do you mind if I rebase your gitweb stack off of the local patch change we have and add a depends on to the upstream change?23:36
clarkbI want to see if that will work23:36
fungirebase at will, and thanks again!23:37
fungii'm interested to see as well23:37
fungisounds like a fun experiment in cross-source dependencies23:37
opendevreviewClark Boylan proposed opendev/system-config master: Drop gitweb dependencies
opendevreviewClark Boylan proposed opendev/system-config master: Fix mixed spaces and hard tabs in Gerrit config
opendevreviewClark Boylan proposed opendev/system-config master: Use Gitea for Gerrit's code browser URLs
opendevreviewClark Boylan proposed opendev/system-config master: DNM: Fail our Gerrit testing for an autohold
clarkbI'll put a hold on the job for 825396 too23:40
clarkbnice the zuul status for that stack even renders the gerrit upsteam change into the list23:40
clarkband gives you a link to it too23:40
clarkbok hold is in place23:42
opendevreviewMerged openstack/diskimage-builder master: Add debian-bullseye-arm64 build test
opendevreviewMerged openstack/diskimage-builder master: debian-minimal: remove old testing targets

Generated by 2.17.3 by Marius Gedminas - find it at!