Wednesday, 2022-06-29

fungiso in summary, main.yaml updates are still working because ansible is able to smart-reconfigure the scheduler on zuul02, and in theory this will clear up on its own over the coming weekend with the weekly restart now fixed00:00
opendevreviewMerged opendev/system-config master: Remove Gerrit 3.4 jobs
opendevreviewMerged opendev/system-config master: Add Gerrit 3.6 jobs
opendevreviewMerged opendev/system-config master: Add Gerrit 3.5 to 3.6 upgrade testing
opendevreviewOpenStack Proposal Bot proposed openstack/project-config master: Normalize projects.yaml
opendevreviewIan Wienand proposed opendev/grafyaml master: Fix datasource UID
opendevreviewIan Wienand proposed openstack/project-config master: grafana: Add infra deployment overview
opendevreviewIan Wienand proposed openstack/project-config master: grafana: add helper script
opendevreviewIan Wienand proposed openstack/project-config master: grafana: Add infra deployment overview
opendevreviewIan Wienand proposed openstack/project-config master: grafana: add helper script
opendevreviewIan Wienand proposed opendev/grafyaml master: [dnm] testing screenshot of .json panel
ianwit is still churning through the xfilesfactor changes05:00
*** akahat|out is now known as akahat|ruck05:15
fricklerI unpaused gentoo builds now since we seem to be running nodepool with the dib release that should fix it05:18
opendevreviewIan Wienand proposed openstack/project-config master: grafana: add helper script
opendevreviewIan Wienand proposed openstack/project-config master: grafana: Add infra deployment overview
opendevreviewIan Wienand proposed openstack/project-config master: grafana: add helper script
opendevreviewIan Wienand proposed opendev/grafyaml master: Use fixed datasource UID
opendevreviewIan Wienand proposed opendev/grafyaml master: [dnm] testing screenshot of .json panel
fricklerianw: c9s is now starting to show build time data, which I think is a success of your fix, but the build status is now "32"? 06:22
ianwyeah, the 32 i think a result of scaling issues.  it should 0 or 1, but can get aggregated to somewhere in between, and there's a x100 happening06:45
ianwi'm waving hands i forget exactly06:46
ianwanyway the .wsp file update is still happening, and it will take a few days for that to percolate out i think, as it rolls up the data into the larger buckets06:46
ianw0/1 data doesn't represent well.  it's a bit of a hack to try and find the last value and continue it over to show the current status06:49
*** jpena|off is now known as jpena07:39
opendevreviewMerged openstack/project-config master: Normalize projects.yaml
*** rlandy|out is now known as rlandy09:33
akahat|ruckfrickler, hey o/09:47
akahat|ruckfrickler, we have gate blocker could you put it on the top of gate?
akahat|rucktop of tripleo queue *09:48
*** dviroel|out is now known as dviroel11:20
fungiakahat|ruck: since it looks like that change is still way at the back of you're gate queue, i'll assume your request remains relevant. promoting 848075,1 to the front of the tripleo gate queue now11:40
fungier, at the back of your gate queue11:41
fungiand done11:42
rlandyfungi: yes - thank you so much!11:54
fungino problem11:55
akahat|ruckfungi, thank you :)11:58
jrosser_i'm trying to run in it's local test mode but failing like this
jrosser_i understand what the error is saying, but what to do about it is less clear to me12:03
fungican you install ansible.module_utils.zuul_jobs into the venv?12:20
jrosser_fungi: i've tried pip install -e . in the root of the repo12:35
*** pojadhav is now known as pojadhav|afk12:52
fungijrosser_: i'm starting to suspect that may have broken stand-alone running of the script13:33
jrosser_fungi: right - i asked someone here much more python aware than me to also take a look, and we're both really confused about the state of the code as it stands13:34
fungijrosser_: i think the except ImportError may also need to catch newer ModuleNotFoundError13:35
jrosser_for now i think i should be able to run it as an ansible module, which is a little more awkward to add debugging to13:35
fungimaybe try that first?13:35
jrosser_really nice feature though to be able to dual-stack ansible modules and CLI like that13:35
fungijrosser_: or try with python 3.7 if you have it handy?13:37
fungilooks like ModuleNotFound error is new in 3.813:38
fungii think the try block doesn't fail when run under zuul because the module has been preinstalled into ansible's namespace13:39
fungiso the except has bitrotted in the face of newer python interpreters13:40
fungimmm, looking back through the git history for another of my projects which does similar try/except on importing, some cases may have raised ModuleNotFoundError as far back as python 3.6, so maybe this hasn't even worked after 3.513:42
fungimaking it backward-compatible with 3.5 gets a bit tricky. what i did in the past was to if not hasattr(__builtins__, 'ModuleNotFoundError'): ModuleNotFoundError = ImportError13:43
fungiand then forward-compat with except ModuleNotFoundError13:44
fungimaybe a bit hacky, there are probably cleaner approaches13:44
fungijrosser_: anyway, i'm increasingly certain that's what's going on13:46
jrosser_thanks for taking a look - i'll use it as an ansible module13:46
fungithough given the length of bitrot for that, i wouldn't be surprised if there are other problems lurking once the except is modernized13:47
clarkbfungi: jrosser_: initially the swift upload stuff was all in its own module and I bet the cli tooling worked then. But then google and s3 and so on were added and things were refactored to split common code out. I wonder if the cli portion was just never exercised after that refactor. It is probably worth asking in the zuul matrix room to see if anyone there knows how to make it work13:56
fungiright, the change i referenced is what refactored those to centralize the logic into upload_utils13:57
fungiand it tried to add relative imports as a fallback when the role isn't installed under ansible, but that's where things are breaking14:01
corvusnone of that is bitrotted -- i use the cli all the time14:01
corvushere's the magic incantation:
clarkbah running it out of a dir that allows the python interpreter to dig down itno the roles and find what it needs14:02
* jrosser_ makes patch14:03
*** dasm|afk is now known as dasm14:03
fungiooh, helpful code comment! would be good to copy that into the other scripts14:04
fungithanks corvus!14:04
opendevreviewJonathan Rosser proposed zuul/zuul-jobs master: Add cli running example for s3 and swift log uploaders
fungistill struggling with working out how to call that from a venv with the dependencies installed14:09
fungiokay, i can do it with a venv outside of tox14:11
fungicd roles14:11
fungipython3 -m venv foo14:11
fungifoo/bin/pip install -r ../test-requirements.txt14:12
fungifoo/bin/python3 -m upload-logs-base.library.zuul_swift_upload --help14:12
fungibut trying to `tox -e venv -- python -m upload-logs-base.library.zuul_swift_upload --help` with roles as my cwd as a short-cut seems to be second-guessing the import path14:14
fungiprobably a tox option or two in a custom testenv for these would work14:15
*** pojadhav|afk is now known as pojadhav14:35
clarkbinfra-root I'm sanity checking and I don't see that project in codesearch results for any other zuul configs implying it is safe to remove from the tenant config and won't create additional errors. Is there anything else you think we should check before removing ap roject from the zuul tenant config like that?15:07
clarkbI suppose that a revert is straightforward and that is why I split the windmill and neutron-classifier removals into two changes as I figure they aren't related to each other so if reverts happen it will be one or the other15:07
clarkbI do think it would be neat to try and land one or both of those changes soon to test some of these assumptions. I have noticed that some openstack projects are starting to push changes up to fix the queue: thing too15:08
corvusclarkb: you could run zuul validate-tenants on it, but i probably wouldn't bother.  just throwing it out there as an option.15:11
fungiahh, okay, i don't need a special testenv definition in tox, this works with the general testenv:venv...15:11
fungitox -e venv -- python -m roles.upload-logs-base.library.zuul_swift_upload --help15:11
clarkbcorvus: ya that seems like overkill (and may haev a bunch of false positives from the other existing errors?)15:12
*** dviroel is now known as dviroel|lunch15:24
*** marios is now known as marios|out15:48
*** rlandy is now known as rlandy|biab15:51
opendevreviewJonathan Rosser proposed zuul/zuul-jobs master: Create the log container and show the url before uploading logs
*** rlandy|biab is now known as rlandy16:27
*** jpena is now known as jpena|off16:34
*** dviroel|lunch is now known as dviroel16:37
opendevreviewJonathan Rosser proposed zuul/zuul-jobs master: Create the log container and show the url before uploading logs
opendevreviewJulia Kreger proposed openstack/diskimage-builder master: DNM: Network Manager logging to Trace for Debugging
*** undefined_ is now known as Guest368417:55
*** Guest3684 is now known as rcastillo_17:55
*** lbragstad1 is now known as lbragstad17:57
*** rcastillo_ is now known as rcastillo17:57
opendevreviewClark Boylan proposed opendev/system-config master: WIP Update to Gitea 1.17.0-rc1
opendevreviewClark Boylan proposed opendev/system-config master: Move gitea partial clone test
clarkbinfra-root I added the breaking change list from the gitea changelog to that commit message and went through each one and left my notes on them. There is one outstanding item that I believe needs testing via a held node which I'll set up if those two changes end up happy via a new DNM failing change18:29
clarkbNote this is only the breaking changes list and I haven't gotten through the rest of the change log18:29
clarkbalso 848174 is reviewable and hopefully mergeable now. It cleans up a testing thing I noticed when adding more testing to this for 1.17.018:30
clarkbThis release is a big one compared to some of the previous ones18:31
opendevreviewClark Boylan proposed opendev/system-config master: DNM intentional failure to hold a node
clarkbI've just put a hold in place for system-config-run-gitea against ^19:56
*** dviroel is now known as dviroel|biab20:30
clarkbhttps:// shows me the favicon I expect. I think I understand the change better after being able to reread the code change20:44
clarkbbasically they used the logo.svg as the favicon.svg and then fell back to favicon.png. But they want to support users having a different logo.svg and favicon.svg so now you must have a separate favicon.svg too. Which we already have20:44
clarkbI'll probably delete that held node today if anyone else wants to check it really quickly. I'll update the parent change that is collecting all this info about the upgrade with this information too20:45
opendevreviewClark Boylan proposed opendev/system-config master: WIP Update to Gitea 1.17.0-rc1
clarkbPTG planning is starting to pick up. I expect that fungi and myself will be there. I don't think ianw or frickler are planning to go.20:50
clarkbcorvus: do you know if you intend on being there at this point? and if so do you think it makes esnse to try and schedule time for opendev if it is the three of us?20:51
clarkbThis isn't urgent but thought I'd start asking the ptg questions in case we do want to plan something.20:53
*** timburke__ is now known as timburke21:01
clarkbneat gitea added a healthcheck
corvusclarkb: thanks; i hadn't realized that was so imminent.  i don't know at this point; i'll try to start figuring that out.21:01
clarkbpossible we could start using that to check the server is up doing rolling restarts and other operations, but it seems basic enough now that it probably isn't too different from what we are alread doing21:02
clarkbcorvus: it is 4 months away which isn't super close but also not super far away21:02
corvusfun fact: (almost) everyone uses healthz because google did.  google did because they already had a "/health"21:03
clarkbgitea 1.17 will add support for docker registries too21:07
clarkb(and python packages and php and maven and so on)21:08
clarkbat first I thought it was integrating with the upstream package repos but no it seems to be the actual repo in gitea21:08
clarkboooh the new go to file feature is actually pretty neat. Try it at
clarkbok I've now skimmed much more of the changelog and I think we're probably in pretty good shape with the WIP change. Of course that can change when the final release happens21:18
clarkbBut I suspect the bulk of the work there is done now and we can do final validation after the final release21:18
*** timburke is now known as timburke__21:39
*** timburke__ is now known as timburke21:40
ianw874687 stats later the xfilefactor resize is still running on graphite21:56
clarkb874687 stats later ?22:04
*** dviroel|biab is now known as dviroel22:09
ianw.wsp files it's updated ... it's still going.  i don't know how many files there actually are22:12
clarkboh I see22:15
*** dasm is now known as dasm|off22:15
*** rlandy is now known as rlandy|bbl22:25
*** dviroel is now known as dviroel|out22:28
opendevreviewChris Stone proposed openstack/diskimage-builder master: Removing old grub removal step
fungisetuptools 63.0.0 is looming, 63.0.0b1 is on pypi if anyone wants to test things with it22:35
clarkbI can't find a changelog for that version just the previous ones22:37
clarkbeven checking out the beta tag it doesn't have changes listed22:37
funginotably for me, it supports editable/develop installs when used as a pep 517 build backend22:38
fungiyeah, i think they write their changelogs just before actual releases22:38
fungiit's mostly been pushed so folks can test pep 660 support:
ianwi guess touching the .wsp files has also expanded their lifespan, as we do a cleanup of anything that hasn't updated in ~9 months i think.  i don't think that's going to blow things out though, we're < half full 1007G  399G  558G  42% /opt23:43

Generated by 2.17.3 by Marius Gedminas - find it at!