Tuesday, 2022-12-20

opendevreviewClark Boylan proposed zuul/zuul-jobs master: Add nox role and some simple jobs  https://review.opendev.org/c/zuul/zuul-jobs/+/86800200:17
opendevreviewClark Boylan proposed zuul/zuul-jobs master: Add inline comment support to nox  https://review.opendev.org/c/zuul/zuul-jobs/+/86813400:17
*** rlandy|bbl is now known as rlandy00:17
ianw... http://storage.bhs.cloud.ovh.net/v1/AUTH_dcaab5e32b234d56b626f72581e3644c/zuul_opendev_logs_5fc/868054/3/check/system-config-run-review-3.6/5fccf0c/bridge99.opendev.org/screenshots/gerrit-main-page.png00:35
ianwthis *does* show the RP column00:35
clarkbianw: whats the difference?00:36
ianwfrom yesterday, doesn't00:36
clarkboh heh did gerrit change something?00:36
ianwno i don't think so00:36
ianwi think with that one, we have 3 changes and i set the votes to -1, +1 and +2 ... none of them have a zero vote00:36
ianwi wonder if its something to do with *all* changes having to have a vote for the column to show?  a thread to pull on00:37
ianwon the change before, i think i had a -2, 0 and +2 vote on the three test changes00:38
ianwok i think i replicated it on
ianwit seems the column only appears if a change has a negative vote00:43
ianwinteresting.  there i've setup "review priority" which goes from -1 -> +2.  i wonder if it's negative, or "max negative" -- i.e. if i change that to have -2 votes, will a -1 make the column appear00:47
ianw... ok it's a "max negative" that makes it appear.  i.e. i added -2 votes ^, and a -1 vote does not make the RP column appear, but a -2 does00:49
ianwi feel like the old behviour was the column appears if any vote is made00:50
ianwknowing that ... i think https://gerrit.googlesource.com/gerrit/+/refs/heads/master/polygerrit-ui/app/utils/label-util.ts is where it's all computed01:15
*** rlandy is now known as rlandy|out01:30
opendevreviewClark Boylan proposed zuul/zuul-jobs master: Add nox role and some simple jobs  https://review.opendev.org/c/zuul/zuul-jobs/+/86800202:09
opendevreviewClark Boylan proposed zuul/zuul-jobs master: Add inline comment support to nox  https://review.opendev.org/c/zuul/zuul-jobs/+/86813402:09
opendevreviewClark Boylan proposed zuul/zuul-jobs master: Add nox role and some simple jobs  https://review.opendev.org/c/zuul/zuul-jobs/+/86800202:41
opendevreviewClark Boylan proposed zuul/zuul-jobs master: Add inline comment support to nox  https://review.opendev.org/c/zuul/zuul-jobs/+/86813402:41
*** yadnesh|away is now known as yadnesh04:10
ianwso i know generally what's going on -- the thing that lays out the changes is looking at the change and essentially looking at the .submit_requirements field, and showing a column for each of those05:20
ianwi think this is a implementation detail of the "anyWithBlock" config item (lowest vote blocks change) -- only if the change has a lowest vote for one of our labels does it's .submit_requirements field have an entry 05:21
ianw  "name": "Code-Review",05:22
ianw  "status": "UNSATISFIED",05:22
ianw  "is_legacy": true,05:22
ianw  "submittability_expression_result": {05:22
ianw    "expression": "label:Code-Review=MAX -label:Code-Review=MIN",05:22
ianw    "fulfilled": false,05:22
ianw    "passing_atoms": [],05:23
ianw    "failing_atoms": [05:23
ianw      "label:Code-Review=MAX",05:23
ianw      "-label:Code-Review=MIN"05:23
ianw    ]05:23
ianw  }05:23
ianwoops, didn't mean to paste that ... but anyway ... that is the submit_requirements object in the change05:23
ianwok, is clearly what you're supposed to do.  use submitRequirementColumns to statically set that column as showing up05:54
ianwthe problem is that doesn't really work for us that well where projects may or may not use this, or may have their own versions of tags.  05:55
ianwanyway, i've got enough insight now to discuss this properly with upstream.  05:56
ianwthis is lightly discussed in https://groups.google.com/g/repo-discuss/c/9QM7d1Cw2tM ... but doesn't quite cover our usage05:56
*** pojadhav- is now known as pojadhav|rover06:04
*** akahat|ruck is now known as akahat|ruck|lunch09:42
*** dviroel|out is now known as dviroel10:58
*** akahat|ruck|lunch is now known as akahat|ruck11:05
*** rlandy|out is now known as rlandy11:19
*** yadnesh is now known as yadnesh|away12:39
fungiyeah, that seems to be a site-wide list of which labels to show, while we want to  get the dynamic inclusion back13:12
*** dasm|off is now known as dasm13:43
*** cloudnull5 is now known as cloudnull14:31
*** jpena|off is now known as jpena14:40
*** dviroel is now known as dviroel|lunch15:43
fungiokay, i've managed to confirm on the held node that `docker exec mailman-web ./manage.py runjobs hourly` does indeed exit nonzero (137) when SITE_ID=0 in the settings.py and not when i switch it to SITE_ID=1, but then rerunning it succeeds when switching it back to SITE_ID=0 after that16:06
fungii wonder if simple lineinfile tasks to undo/redo that edit around the manage.py call would suffice16:06
opendevreviewJeremy Stanley proposed opendev/zone-opendev.org master: Dummy mailman hostname to house the list of lists  https://review.opendev.org/c/opendev/zone-opendev.org/+/86798116:14
opendevreviewJeremy Stanley proposed opendev/system-config master: Use magic domain guessing in Mailman 3  https://review.opendev.org/c/opendev/system-config/+/86798716:14
opendevreviewJeremy Stanley proposed opendev/system-config master: DNM force mm3 failure to hold the node  https://review.opendev.org/c/opendev/system-config/+/85529216:14
fungiokay, new autohold in place for 855292 on the assumption that 867987 will pass this time. also 867981 was a rebase to resolve the serial conflict from the iweb mirror server cleanup16:19
clarkbfungi: and you're still aiming to test that on he hold before landing anything?16:22
fungiyeah, though i did test that the temporary edit should solve the warning from the faq about initial migrations run breaking (what i think the job failure on the previous revision demonstrated too)16:22
fungii wonder if we should be running migrations earlier in the job though, like as part of the server deploy16:23
fungisince right now it's only getting exercised in the test16:23
clarkbare we automating any migrations?16:24
fungithese are database migrations16:24
fungiwhich run from the hourly cronjobs on the server16:24
clarkboh db migrations. They are run when the service starts16:25
clarkbpretty sure if you look at the service logs part of the startup is the migrations16:25
clarkband part of what we wait for on setting up the admin user is sufficient db content16:25
fungiexcept it's only `./manage.py runjobs hourly` which seems to fail with a migrations problem, container startup works fine16:25
clarkbya that is odd. Maybe they don't run all of the migrations on startup just a minimal set16:26
clarkb(an interesting design feature...)16:26
fungiso if we stand up a new server with SITE_ID=0 then the hourly cronjobs in the mailman-web container will break until we run them at least once with SITE_ID=1 temporarily16:27
funginot sure how exactly to encode that into ansible for actual server deployment16:27
fungi(for the test it seems easy enough)16:28
clarkbI think you can start the deployment with site id = 1. Then run the hourly cron manually if site id = 1 and if that succeeds flip side id to 016:28
clarkbthen you won't run the migrations explicitly in the future because you've flipped the switch on them16:28
fungioh, interesting approach. so pseudocode "if SITE_ID=1 exec hourly cron and set SITE_ID=0"16:29
fungihow do we reconcile that with deploying new copies of the settings.py file though?16:30
fungisince that would undo the SITE_ID=016:30
clarkbwe would probably need to template the file with a default value of 1, but then set SITE_ID to the value that already exists in the file maybe? You're right that this is clunky (and largely a side effect of mailman not being able to run under its expected config from the start)16:31
clarkbor use versions of the file and select the appropriate file version based on where you are in that conditional logic16:32
clarkbthis might be simpler to implement in ansible but result in copying extra content around16:32
fungiwe could drop a flagfile to track whether the initial exec has happened and if that file isn't present then switch to SITE_ID=1 temporarily and run the exec and switch back to SITE_ID=0 to match the copy of the file we deploy?16:33
fungiworst case someone blows away the flagfile and ansible reruns the cron once16:34
fungithat avoids templating or having multiple copies of the config16:34
clarkbya that might work16:34
clarkband it might be useful to be able to delete that file and have it rerun16:35
fungii'll try to work on that if i get another break between meetings today16:35
opendevreviewClark Boylan proposed opendev/bindep master: Use nox  https://review.opendev.org/c/opendev/bindep/+/86800416:37
*** marios is now known as marios|out16:46
*** dviroel|lunch is now known as dviroel16:59
opendevreviewJames E. Blair proposed opendev/base-jobs master: Add nox-docs base jobs  https://review.opendev.org/c/opendev/base-jobs/+/86822617:08
opendevreviewJames E. Blair proposed opendev/base-jobs master: Add nox-docs base jobs  https://review.opendev.org/c/opendev/base-jobs/+/86822617:15
opendevreviewJames E. Blair proposed opendev/base-jobs master: Add nox-docs base jobs  https://review.opendev.org/c/opendev/base-jobs/+/86822617:17
*** jpena is now known as jpena|off17:20
fungiopendev holiday party is happening in https://meetpad.opendev.org/opendev-holiday-party and #opendev-meeting and on https://boardgamearena.com/19:07
fungilatest revision to 867987 still didn't work, i wonder if we need to down and up the mailman-web container with each change to settings.py20:30
fungii hadn't thought about it, but i did that when i was testing on the held node so maybe that's why it worked for me20:30
Clark[m]Or maybe hup it at least20:31
ianwfungi/clarkb : https://etherpad.opendev.org/p/-MyiqZX4dijrluhkNuXx is my draft email about the column display.  it does tie in with the submitrequirements, but one thing at a time :)  i'm having trouble ending it with a concrete suggestion21:17
rgunasekaranHello All,21:19
rgunasekarannew to this group & the community so bear with me while i get my footing 21:19
rgunasekaranI want to contribute, can you help me picking out something to work on? please suggest on accurate successful documentation for setting up the environment and start contributing!21:19
clarkbrgunasekaran: is there anything in particular that interests you? Right now the main things we've got going on are a few conversions of services to containers (from puppet), sorting out Gerrit ugprades (what ianw is working on above), dealing with the recent tox v4 release (it isn't backward compatible), and updating our existing containers to newer versions of python.21:23
clarkbOh and finally some server upgrades from ubuntu bionic to jammy21:23
clarkbianw: couple of notes on your etherpad draft but lgtm21:25
rgunasekaran@clarkb: i am interested to work with containers & kubernetes! please let me know how i can contribute for the same!21:27
clarkbwe don't currently run any kubernetes. All of our containers are run by docker-compose managed by ansible. Most of that can be found in https://opendev.org/opendev/system-config under the docker/ and playbooks/roles/ dirs21:28
clarkbOne example activity that would be helpful is updating our images like the refstack image (https://opendev.org/opendev/system-config/src/branch/master/docker/refstack/Dockerfile) to python3.10 or 3.11 depending on compatibility. With the eventual goal that if we get everything off of python3.9 and can stop building those images21:30
ianwclarkb: thanks!  i'll come back it and let it percolate, see if i have any other ideas now i understand what's happening21:30
clarkbhttps://codesearch.opendev.org/?q=opendevorg%2Fpython-builder%3A3.9-bullseye&i=nope&literal=nope&files=&excludeFiles=&repos= shows you what is still on 3.9 and if you look at the git history for those files you'll see their evolution from 3.6 -> 3.9 or so21:30
rgunasekaranclarkb: okay i can start contribute for containers run by docker-compose managed by ansible21:33
clarkbrgunasekaran: another good place to start to get familiar with things is to look at the existing changes that have been proposed21:34
clarkbrgunasekaran: https://review.opendev.org/q/status:open+project:opendev/system-config for example21:35
clarkbthat will show you what others are currently work on to get a sense for what changes are happening21:35
rgunasekaransure @clarkb i am there on the opendev changes 21:38
rgunasekaranlooking at the open changes!21:38
*** dviroel is now known as dviroel|out21:42
clarkbinfra-root: I'm working on collecting possible content for the OpenDev Foundation Annual Report section. Currently plan on calling out the Gerrit upgrades, work with Zuul to get to fully redundant deployment and automated upgrades, (slow) transition to Jammy, and python3.10/3.11 bootstrapping. I'm sure there is moreand I still need to look at git logs and my historical notes file for21:55
clarkbideas. Anything you'd like to see get called out?21:55
clarkboh the centos streaming happened21:56
clarkbWe fixed the Gerrit ssh rsa problem21:57
clarkbWe shutdown the ELK system21:57
clarkbAnd ethercalc21:57
ianwansible upgrades across zuul maybe?21:57
clarkbThe bastion updates should be mentioned too21:58
clarkbWe deployed a keycloak and mm321:59
clarkbThe jitsi meet upgrades22:00
ianwnot sure how to phrase it but things that we noticed/did that went back into zuul upstream; e.g. streaming console, quite a few UI/PF4 upgrade things, other bugs i've forgotten22:00
clarkbIn the past I think I've descirbed how we have a feedback loop between zuul and opendev. We dogfood zuul and help make it better. I think we should continue to cal lthat out22:01
fungirevisiting the non-submit-requirement labels in gerrit, i wouldn't be surprised if they come back with the suggestion to use hashtags instead22:01
clarkbfungi: thats a good point. I don't think hashtags get a summary on listings though so may not fully solve the problem. But I'm willing to try it if that is their response22:02
ianwfungi: yeah; although we do have the ones that are setup to block on negative votes too.  with 3.6 they're doing the weird thing of any blocking vote will make the summary column appear, but non-blocking ones won't22:03
fungiin a way, these use cases do seem like overloading the label/vote mechanism for off-label (pun intended) purposes22:03
ianwyeah, that was what i was trying to get at with the "gerrit -> developer" communication, versus "developer -> developer" communication 22:04
*** JasonF is now known as JayF22:39
ianwi've uploaded a ubuntu base image to the new linaro cloud; it sort of seems to work but has no console.  not 100% sure what's going on.  probably best to trying bringup of a mirror node as first thing22:49
ianwi've created some users and projects that should hopefully wokr22:49
opendevreviewIan Wienand proposed opendev/system-config master: Linaro: initial cloud config  https://review.opendev.org/c/opendev/system-config/+/86824723:19
clarkbthat looks good to me. I think you can probabl approve it once the groupvars/hostvars are sorted23:33
ianwthanks, yeah i want to watch the cloud launcher run closely, i may well have stuffed something up on the cloud side23:37
opendevreviewIan Wienand proposed opendev/system-config master: Linaro: initial cloud config  https://review.opendev.org/c/opendev/system-config/+/86824723:41
ianwproject_name/project_id typo.  not sure why we have both23:41
clarkbI think thats a long term holdover from some clouds only using ids and not names?23:54
clarkbits probably slightly faster to use ids too23:55
clarkbbut I like names for human interaction23:55

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