Tuesday, 2023-01-31

opendevreviewIan Wienand proposed zuul/zuul-jobs master: promote-docker-image: some notes on manual replication  https://review.opendev.org/c/zuul/zuul-jobs/+/87223700:13
opendevreviewClark Boylan proposed opendev/system-config master: Switch Gerrit test Verified label function to NoOp  https://review.opendev.org/c/opendev/system-config/+/87223800:23
clarkbI've done some digging into the gerrit source since my emails upstram didn't get any bites. I think that we may actually need to do the explicit NoOP00:23
clarkbI'm about to put the meeting ageand together. Let me know if there is anything I should add/remove (I'm adding the git udpates and removing gitea 1.18, linaro cloud, and pruning backups). ianw  let me know if I should keep linaro on there but I Think things are running ?00:27
ianwyeah i think it's going ok.  we might like to discuss how much we'd like to manage the underlying system00:28
clarkboh good idea. I'll keep it on the agenda then00:28
opendevreviewIan Wienand proposed openstack/project-config master: Remove manila-tempest-dsvm-hdfs  https://review.opendev.org/c/openstack/project-config/+/87223900:34
opendevreviewIan Wienand proposed openstack/project-config master: Remove manila-tempest-dsvm-hdfs  https://review.opendev.org/c/openstack/project-config/+/87223900:44
clarkbok agenda is updated. I'll give it a few minutes before sending out00:53
clarkband sent01:03
*** rlandy|bbl is now known as rlandy01:36
*** rlandy is now known as rlandy|out01:39
*** dasm is now known as dasm|off01:44
*** yadnesh|away is now known as yadnesh04:41
opendevreviewIan Wienand proposed zuul/zuul-jobs master: build-docker-image: fix change prefix  https://review.opendev.org/c/zuul/zuul-jobs/+/87225806:20
*** yadnesh is now known as yadnesh|afk07:57
*** jpena|off is now known as jpena08:02
*** ysandeep__ is now known as ysandeep|lunch09:03
opendevreviewAndreas Jaeger proposed zuul/zuul-jobs master: promote-docker-image: some notes on manual replication  https://review.opendev.org/c/zuul/zuul-jobs/+/87223709:04
opendevreviewMerged opendev/infra-manual master: General updates for the Test Environment chapter  https://review.opendev.org/c/opendev/infra-manual/+/87220109:11
*** yadnesh|afk is now known as yadnesh09:25
*** ysandeep|lunch is now known as ysandeep10:50
*** rlandy|out is now known as rlandy11:15
opendevreviewRodolfo Alonso proposed openstack/project-config master: Add github sqlalchemy/sqlalchemy to the project list  https://review.opendev.org/c/openstack/project-config/+/87227511:28
opendevreviewRodolfo Alonso proposed openstack/project-config master: Add github sqlalchemy/sqlalchemy to the project list  https://review.opendev.org/c/openstack/project-config/+/87227511:28
*** dviroel|out is now known as dviroel|rover11:29
*** ysandeep is now known as ysandeep|brb11:49
*** tkajinam is now known as Guest300312:02
*** ysandeep|brb is now known as ysandeep12:23
opendevreviewVishal Manchanda proposed openstack/project-config master: Step 1: End Project Gating, Retire xstatic-font-awesome  https://review.opendev.org/c/openstack/project-config/+/87170312:48
fungipip 23.0 is out today: https://pip.pypa.io/en/stable/news/#v23-013:19
rpittauhello everyone! The ironic project has hit a little bump on the road, our CI is dependent from the virtualpdu project (now in x, but before was in openstack) which is unfortunately not very Python 3 friendly, and we can't move to jammy because of that.13:24
rpittauWe've prepared a couple of patches to fix that, and contacted the project cores, but we got no answer so far after a couple of weeks.13:24
rpittauWe're ready to fork or even adopt virtualpdu under the ironic umbrella and maintain it, just wanted to see what the procedure would be in case we go down that road, thanks!13:24
fungirpittau: if the current maintainers can't be reached for permission to take over maintenance, you'd need to fork it. If you wanted to fork into the openstack namespace as a deliverable of ironic, submit a project creation change like usual and make https://opendev.org/x/virtualpdu the upstream import url13:43
fungiif you want to release it on pypi, you'll probably want a new name for it though, since pypi doesn't have namespacing13:43
rpittauthanks fungi13:45
fungirpittau: if you can get permission to take it over, then ideally have one of them add you to the group(s) for it gerrit. you can optionally work with us to move it into the openstack namespace once you're a maintainer13:47
fungis/for it gerrit/for it in gerrit/13:47
rpittaufungi: that would probably be preferable, need to talk to the rest of the ironic cores13:47
fungiavoiding forking would be preferable, but someone would need to be able to get in touch with the current maintainers for that13:48
rpittauI'll give another try13:49
rpittauthe huge problem we have now is that we can't move all ofthe CI jobs to jammy, so we need to find a workaround quickly13:49
rpittauJayF, dtantsur, TheJulia FYI ^^13:56
fungihow specifically is it breaking on jammy? maybe there's another short-term workaround13:59
dtantsurfungi: jammy has no python 2, I assume14:01
dtantsuras a short term workaround, we can prepare a container based on whatever ubuntu version it supports14:02
dtantsurbut ideally we shouldn't rely on a dead project14:02
fungioh, it was never made to work with python3?14:02
rpittaufocal is supported, the problem starts with jammy and python 3.10, I think14:02
rpittauthe higher version supported is python 3.514:02
dtantsurahh, sorry. but yeah, the situation is similar.14:03
fungirpittau: dtantsur: not great as a long-term solution obviously, but maybe https://launchpad.net/~deadsnakes/+archive/ubuntu/ppa can help temporarily?14:04
fungithey oldest python they have for jammy is 3.7 but perhaps that's old enough14:05
rpittauit might work, if that works with python 3.8 we can probably steal some more time14:07
fungipackages from thereshould be installable alongside the default python provided in jammy, so you can invoke python3.7 or whatever directly to install virtualpdu (i assume you're running it as an application and interacting with it that way, not importing it into ironic code)14:08
rpittauyep, correct14:09
fungimgagne: do you happen to be in touch with the (former?) inap people who worked on virtualpdu? looking through the virtualpdu-core group in gerrit i see a lot of internap addresses14:14
*** dasm|off is now known as dasm14:28
*** ysandeep is now known as ysandeep|out14:33
Clark[m]I think jammy does have python214:36
Clark[m]https://packages.ubuntu.com/jammy/python2 is your shorter term workaround14:37
fungiahh, yeah14:38
rpittauClark[m]: thanks, I'll keep that as the very last option, trying to make it work with python 3.8 now14:41
Clark[m]Where is python 3.8 coming from? Chances are distro python 2.7 for a project that never really did python 3 is going to be your best bet14:42
rpittauaccording to the setup.cfg virtualpdu should support up to python 3.414:43
rpittaubut it was tested with python 3.514:43
clarkbright but ppa python 3.8 on jammy is going to be less reliable than installing python2 out of a local mirror. Also python2 on jammy isn't going anywhere but I'm not sure what that PPAs rule for python versions that eol is14:50
rpittauerrr this is really a mess.... it does not work with py 3.8 and 3.7 (just tried) and also we can't use python 2 cause libvirt-python needs python >= 3.5 :/15:05
fungihow was it working on focal? older libvirt-python?15:05
fungii guess you could cap libvirt-python in the virtualpdu python version15:06
rpittauI'm thinking that15:06
fungii guess the pip version being used with python2 doesn't support the requires_python metadata15:07
fungior libvirt-python doesn't set it correctly15:08
*** dviroel|rover is now known as dviroel|rover|lunch15:08
clarkbjammy pip should do that15:08
clarkbits the python3 pip installing to python2 I would expect15:08
fungidepends on whether you invoke python2 -m pip or something15:08
clarkbbut ya looking at pypi they don't seem to set requires python15:08
fungii suppose we're lucky libvirt started publishing their python bindings to pypi at all15:09
clarkbI think you need libvirt-python<615:09
rpittaudoing it now15:10
rpittauand of course libvirt bindings don't work15:11
rpittauyep, it fails building libvirt-python15:14
clarkbhrm I wonder if that older library isn't new enough to talk to jammy's libvirt then?15:14
fungioh, right, i think they need to ~match15:17
fungiyeah, that's quite the mess15:17
clarkbI think the python lib needs to be >= to the libvirt install. Newer python lib is fine. Not that that helps in this case15:18
rpittauforgot how strict libvirt is, but it makes sense15:19
fungiso, yeah, python docker container is starting to sound like the easiest alternative to forking at this point15:19
rpittauunless we find another solution I guess we go for a container as a temp workaround15:20
rpittaubut in the long term we really need to be able to fix virtualpdu for py315:20
fungiyou could "fork" it in your job by checking out the git repo, applying a patch/fetching a gerrit change to make it work with python3.10, and then install from that local path15:21
fungiwe do that as an ugly workaround for some of our services in opendev when we need to deploy fixes which haven't merged upstream due to unresponsive maintainers15:22
rpittauI also thought about that, but it requires a lot of patching15:22
rpittauI'd rather fork it entirely and be done with that15:22
rpittauI'll try bumping the mail and see if cores reply15:23
clarkbgit am, or patch make it easy especially if you're rwriting the fixes anway15:23
clarkbbut ya its been useful for sorting out gerrit patches in particular before we push them upstream15:24
JayFThe patches already exist for py3 compatability, right? 15:26
JayFWe just have nobody empowered to land them15:26
rpittauJayF: not entirely15:26
rpittauJayF: I was trying to get the entire project in shape, removing old project config and setting up correct CI jobs15:28
rpittauit takes some effort15:28
JayFyeah I'd imagine so15:28
JayFmy prefererence is, lets just say I'll be thinking of names lol 15:28
JayFwith the hope that the existing group will return an email 15:29
rpittaufungi, clarkb, in the meantime do you think could it be possible to move forward with the patches impacting the project config? https://review.opendev.org/c/openstack/project-config/+/869979 and https://review.opendev.org/c/openstack/openstack-zuul-jobs/+/87051915:47
clarkbI think that is ok. I didn't approve in case fungi had an objection15:48
rpittaugreat, thanks!15:48
fungiyeah, i approved it15:52
fungii'll recheck and approve the second once the first applies15:52
rpittauthanks both :)15:54
fungioh, somehow we've broken the manila-tempest-dsvm-hdfs definition somewhere15:54
fungithat's preventing us from merging further job configuration changes in project-config15:55
clarkbfungi: I think I saw ianw pushing changes to fix that.15:55
clarkbI didn't realize that was why/me looks for them15:55
clarkbI've approved ^ which should allow the other change to land once it merges15:57
fungia few other project-config changes seem to have hit the same error15:57
opendevreviewMerged openstack/project-config master: Remove manila-tempest-dsvm-hdfs  https://review.opendev.org/c/openstack/project-config/+/87223916:03
*** dviroel|rover|lunch is now known as dviroel|rover16:14
fungirpittau: one of the virtualpdu core reviewers just left a +1 on 86997916:23
*** jpena is now known as jpena|off16:25
rpittaummm he approved also the change in the virtualpdu project itself, although it's not complete, hopefully we'll have a quicker turnaround next iteration16:25
opendevreviewMerged openstack/project-config master: Reduce virtualpdu tests  https://review.opendev.org/c/openstack/project-config/+/86997916:29
*** yadnesh is now known as yadnesh|away16:35
rpittaufungi, clarkb, JayF, sorry to bother you again! We just got an answer from 2 cores of virtualpdu and one of them gave us his blessing to take over virtualpdu, how do we formalize this?16:39
fungirpittau: if they can go to https://review.opendev.org/admin/groups/q/filter:virtualpdu while logged in and add one of you as members to the two groups listed there, that's the first step16:41
rpittaufungi: cool, thanks16:41
clarkbya the best thing is if we don't have to intervene at all and our users can do a graceful handover16:41
fungionce you have control of group membership, if you want to move it officially into openstack then you'll want a governance change and project config changes to rename the repository, and add an item to our weekly meeting agenda here so we can discuss scheduling a rename maintenance window: https://wiki.openstack.org/wiki/Meetings/InfraTeamMeeting#Upcoming_Project_Renames16:42
rpittauok, I guess we'll have to discuss about that with the rest of the ironic community16:43
fungiyep, one step at a time. but at least now you'll have access to fix up your dependency even if it's not technically in openstack16:45
fungiand you have a path to adopting it officially in openstack if the team decides that's the way you want to go16:45
rpittauthat's the most important thing right now :)16:45
fungiwithout needing to fork16:46
JayFrpittau: I'd say we basically decided as a community to take it over if possible16:54
JayFrpittau: so you have my blessing, at least, to move forward. If you want me to poke at any of the project-config or governance changes lmk16:54
opendevreviewJeremy Stanley proposed opendev/system-config master: Separate mailman default hostname  https://review.opendev.org/c/opendev/system-config/+/86798616:56
opendevreviewJeremy Stanley proposed opendev/system-config master: Use magic domain guessing in Mailman 3  https://review.opendev.org/c/opendev/system-config/+/86798716:56
opendevreviewJeremy Stanley proposed opendev/system-config master: DNM force mm3 failure to hold the node  https://review.opendev.org/c/opendev/system-config/+/85529216:56
*** marios is now known as marios|out17:01
opendevreviewJeremy Stanley proposed opendev/system-config master: DNM: Unhide task output  https://review.opendev.org/c/opendev/system-config/+/87233817:31
fungiclarkb: trying to add a container restart into the mm3 deployment process, what's the best way to pause task execution until it's up? just add another task polling the api?18:38
clarkbfungi: "yes" there are some tasks that already do that in the mm3 role after the initial deployment. I've seen that the api can be up before django is full initialized. THough after the initial install I uspect that checking the api endpoint is probably sufficient18:40
fungithanks, i'll see if that's sufficient18:43
opendevreviewJeremy Stanley proposed opendev/system-config master: Use magic domain guessing in Mailman 3  https://review.opendev.org/c/opendev/system-config/+/86798718:43
clarkbI think the issue is that the api doesn't rely on django being there at all18:48
fungiclarkb: also as far as adding sites in django i was pointed to https://docs.djangoproject.com/en/4.1/ref/contrib/sites/18:49
clarkbskimming that there doesn't seem to be an easy mode create a site. But that at least documents the expectations18:50
clarkboh apparently we could manage our own db migrations though?18:51
clarkbthat seems wrong (migrations should manage schemas and necessary data transformations not the data itself)18:51
fungii think https://docs.djangoproject.com/en/4.1/ref/contrib/sites/#enabling-the-sites-framework explains how to create sites? it's a bit opaque though18:53
clarkbya it basically says create a custom db migration to inject your data18:53
clarkbwhich to me is the absolute wrong way to use db migrations18:54
fungihmm, so db migrations as a way to run inserts?18:54
clarkbyup at least that is how I'm reading that section18:54
fungireading some additional articles about django, i think i see why18:56
fungieach django "site" has a dedicated set of database tables, and the migrations are being used to create the tables18:57
fungilooks like django-admin makemigrations has a --scriptable option which might be handy19:01
*** dasm is now known as dasm|off19:33
ianwclarkb: i'm not sure i realised https://hub.docker.com/_/openjdk was deprecated?19:39
clarkbianw: its like half deprecated19:39
clarkbwhich is why I think it wasn't really clear19:39
clarkbbut ya the bits we are using need to not be used anymore19:40
opendevreviewJeremy Stanley proposed opendev/system-config master: DNM force mm3 failure to hold the node  https://review.opendev.org/c/opendev/system-config/+/85529219:50
opendevreviewJeremy Stanley proposed opendev/system-config master: Feature our cloud donors on opendev.org  https://review.opendev.org/c/opendev/system-config/+/86909119:55
*** dviroel|rover is now known as dviroel|uot21:00
fungiianw: clarkb: looks like the debian container images may have updated git now, based on the failure i'm seeing for system-config-build-image-gitea here: https://zuul.opendev.org/t/openstack/build/6f39eb72a9464fa790c43428589201f421:25
fungi"The following packages will be DOWNGRADED: git git-man"21:25
fungiare we clear to approve 872209 now?21:26
clarkbfungi: I think that error is what zuul hit too and the leaf image (eg zuul) is installing git to he patched version and then apt is angry with our local package. However, the base image we pull from hasn't updated yet at last check21:27
clarkbfungi: and ya I think landing 872209 is a good idea as is https://review.opendev.org/c/opendev/system-config/+/87221221:27
fungidone and done21:28
fungitaking a quick break to cook dinner, but can check in on the deployment afterward21:28
clarkbI'm going to need to do a school run in not too long. THe next thing on my todo list was picking up zuul sqla debugging again. But not sure i have time to page all that in before I need to pop out21:33
opendevreviewAde Lee proposed openstack/project-config master: Add base openstack FIPS job  https://review.opendev.org/c/openstack/project-config/+/87222222:20
opendevreviewAde Lee proposed zuul/zuul-jobs master: Add ubuntu to enable-fips role  https://review.opendev.org/c/zuul/zuul-jobs/+/86688122:24
opendevreviewMerged opendev/system-config master: Revert "Update git in gitea images"  https://review.opendev.org/c/opendev/system-config/+/87220922:48
opendevreviewMerged opendev/system-config master: Flip Gerrit image to install distro git  https://review.opendev.org/c/opendev/system-config/+/87221222:55
fungigitea seems to still be working23:08
*** rlandy is now known as rlandy|out23:10
clarkband fetching the image and checking dpkg shows what I expect23:11
ianwclarkb: just sent a response to the gerrit label function default message, after looking a bit ...23:24
ianwit feels perhaps not exactly a bug, but certainly very difficult to use23:25

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