Tuesday, 2023-10-10

opendevreviewMerged openstack/project-config master: Add OpenStack Keystone LDAP charm  https://review.opendev.org/c/openstack/project-config/+/89648612:09
fungi#status log Pruned backup volume on backup02.ca-ymq-1.vexxhost from 93% down to 64% utilization13:28
opendevstatusfungi: finished logging13:28
clarkbthank you for doing that14:46
clarkbfungi: I don't see emails for that?14:47
clarkbare they being filtered into a specific folder now maybe?14:47
fungiclarkb: i moved them into INBOX.services.backup after the prune completed14:49
fungiwe'd been getting the reminders for about a week14:50
clarkbah ok I just haven't been paying enough attention I guess14:50
clarkbI don't check that inbox daily but do try to get to it at least once a week14:50
fungii try to check it every morning, but if there's a lot going on i often don't get to it14:55
fungi#status log Manually refreshed the Debian mirror after cleaning up a stale reprepro lockfile left behind when the Debian 12.2 point release resulted in the process getting killed by the managing script's timeout wrapper15:00
opendevstatusfungi: finished logging15:00
opendevreviewJames E. Blair proposed opendev/system-config master: Add a debug python base image  https://review.opendev.org/c/opendev/system-config/+/89784415:05
clarkbcorvus: ^ I think that change needs to add the jobs to pipelines?15:14
corvusderp :)15:15
clarkbalso nice that the python images make this easy15:15
opendevreviewJames E. Blair proposed opendev/system-config master: Add a debug python base image  https://review.opendev.org/c/opendev/system-config/+/89784415:18
corvusyeah, i spent yesterday doing a couple of iterations to figure this out, and i think this is the way.15:18
corvusthere's a new service called debuginfod, which is cool in that it'll download debug symbols over http on request, so you don't have to bloat your images.  debian runs a server, but of course that only works for debian packages.  since the python images are compiling python in their dockerfile, we really should use the debug symbols from there.15:29
opendevreviewJames E. Blair proposed opendev/system-config master: Add a debug python base image  https://review.opendev.org/c/opendev/system-config/+/89784415:30
clarkbcorvus: ya tumbleweed/suse have a debuginfod server as well which I discovered by accident recently when loading something into gdb15:32
clarkboh it was python too15:32
clarkbfungi: if you get a chance https://review.opendev.org/c/opendev/system-config/+/895522 is something we probably should've combined with the bookworm update16:01
clarkband then there are the lodgeit and gear python3.11 changes to update their container images as well16:01
opendevreviewDmitriy Rabotyagov proposed openstack/diskimage-builder master: Stop creating default user for cloud-init  https://review.opendev.org/c/openstack/diskimage-builder/+/89132216:04
corvusfungi: can you take a look at https://review.opendev.org/c/opendev/system-config/+/897844 ?  i think it's ready to go17:48
corvusclarkb: fungi fyi i have manually verified that i can use it to get a zuul debug image with https://review.opendev.org/c/zuul/zuul/+/89785917:48
corvusbasically, if i pull the image built there, i can get a python backtrace out of gdb17:48
fungioh that's awesome18:23
fungipython 3.13.0a1 scheduled for friday the 13th19:09
fungiauspicious start to a new development cycle19:09
JayFwatching the gentoo python channel when they release late alphas/early betas is like a spectator sport; I'm pretty sure tests on many python packages get run for the first time with python-asserts-enabled in gentoo19:10
JayFget to go read the most interesting bugs19:10
JayFthey are only interesting b/c I'm reading and not discovering them though :D19:11
opendevreviewMerged opendev/system-config master: Add a debug python base image  https://review.opendev.org/c/opendev/system-config/+/89784419:11
tonybsomething thing I'd like get my head around, and I suspect it may be useful for others, is a backlog list and current work items20:01
tonybthe latter is easy enough to pull from the meeting agenda 20:01
tonybbut is the former tracked anywhere?20:02
tonybalso I get the feeling that infra-specs is a little out of date.20:02
fungiit is very out of date, yes20:09
fungiheading out to grab early dinner, then happy to discuss once done20:10
clarkbyes, it has been somethign we have struggled with. I think we all end up with disparate TODO lists in our local note taking apparatus which maybe contributes to it not being shared properly20:18
clarkbmaybe we should do a big pile on session of recording things somewhere (even if it is just in a git repo or etherpad) then using actual content we can determine where a good long term home for it would be?20:19
JayFtonyb: an approach I've started taking for Ironic and TC, for short/medium term items (it's new enough I don't know how well it works); is meeting action items with a delivery time that isn't "next meeting". You touch on them each meeting to get a status update and push it forward another week. 21:04
fungitonyb: just to second (third?) other observations, we mostly have an intuitive consensus/sixth sense for what needs doing the most and each sort of focus on what we can do. our esteemed "service coordinator" sort of loosely keeps track of what's going on, but trying to communicate these better with the rest of the group and/or rekindle the specs list is likely a good idea21:47
fungia lot of the time, it's honestly just "whatever comes up and seems urgent"21:48
fungithough as a result, some larger efforts (mass platform upgrades, mailman v3 migration, sso roll-out, et cetera) sit on the back-burner for way too long21:49
clarkbI'm remembering all the fun jdb is. No readlien support so you can't rerun old commands easily. No tab completion and you ahve to use full classnames for everything. Quickly adds up to a painful experience21:53
clarkbgood news is system.out.Println does emit things in a capturable format when asking bazel(isk) to run the unittests. So I may just end up iterating on it that way21:54
fungisome variation of that is my go-to barbarian debugging technique in basically any language21:55
fungi"just print out whatever you're unsure about"21:56
clarkbalso jdb works well enough that I think I can use it to stick a breakpoint in place at the end of my test so I can examine the on disk state as well. Might give me a hacky way to see the temporary replication log21:56
fungino, wait, better phrasing is "when in doubt, print it out"21:57
clarkbthough you know what I may have just had a bit of a eureka moment21:57
clarkbI think the reason my test is failing is beacuse the plugin thinks it has to continue to retry the things it isn't cleaning up21:57
fungiretry-on-failure algorithms are the bane of debugging21:57
clarkbmy latest update must not cause it to go from "nothing to do" -> fine -> end and instead "nothing to do" -> expected work hence error -> not fine21:57
clarkbyup there are definitely retries occuring. I need to put this down for the day, but this is probably the thread to pull on22:12
clarkbok really have to pop out now to get kid at school. But I haev confirmed there are retries occuring due to "Repository Missing" I've done that becaus when you precreate the repo in the test suite it creates refs/meta/config for you but I don't awnt it there in order to easily confirm that it doesn't replicate. So I have the replication plugin create the repo. Maybe what has me all22:22
clarkbconfused is the handling of missing repos.22:22
clarkbIn this case the repo will never be created so we actually ant to short circuit which should be doable? I'll have to dig in more another time22:22

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