Tuesday, 2021-08-03

opendevreviewIan Wienand proposed zuul/nodepool master: functional-test : add an external connectivity check  https://review.opendev.org/c/zuul/nodepool/+/80330102:17
*** bhavikdbavishi1 is now known as bhavikdbavishi04:48
opendevreviewIan Wienand proposed zuul/nodepool master: [wip] functional-test : add an external connectivity check  https://review.opendev.org/c/zuul/nodepool/+/80330104:53
*** marios is now known as marios|ruck05:50
swestcorvus: all in all lgtm. some thoughts on 80324506:17
swesttobiash: ^06:31
tobiash[m]yepp, I think adding a release note for this makes sense06:31
tobiash[m]we could cleanup the possibly leaked events on the old path in the cleanup worker for a while06:33
abitrolly[m]<fungi "abitrolly: i know there are zuul"> Oh, `bubblewrap` is nice. I basically still can't find a tool that allows me to test run an arbitrary code from GitHub to see if it is alive. I wonder how ready it is..06:47
opendevreviewIan Wienand proposed zuul/nodepool master: [wip] functional-test : add an external connectivity check  https://review.opendev.org/c/zuul/nodepool/+/80330106:48
*** jpena|off is now known as jpena06:54
*** sshnaidm|afk is now known as sshnaidm07:27
*** rpittau|afk is now known as rpittau07:30
opendevreviewMerged zuul/zuul master: Don't use transactions when sharding large events  https://review.opendev.org/c/zuul/zuul/+/80315407:48
opendevreviewMerged zuul/zuul master: Clean up leaked event queue side channel data  https://review.opendev.org/c/zuul/zuul/+/80324508:01
opendevreviewMerged zuul/zuul master: Store build params separate from request  https://review.opendev.org/c/zuul/zuul/+/80315508:01
opendevreviewMerged zuul/zuul master: Cleanup leaked build request parameters  https://review.opendev.org/c/zuul/zuul/+/80326208:01
opendevreviewMerged zuul/zuul master: Move executor api cleanup to apscheduler  https://review.opendev.org/c/zuul/zuul/+/80326408:01
opendevreviewMerged zuul/zuul master: Remove transaction support from sharding API  https://review.opendev.org/c/zuul/zuul/+/80315608:01
opendevreviewMerged zuul/zuul master: Remove redundant get when removing a build request  https://review.opendev.org/c/zuul/zuul/+/80315808:01
opendevreviewMerged zuul/zuul master: Vendor and use fixed Kazoo read/write lock  https://review.opendev.org/c/zuul/zuul/+/80316608:01
opendevreviewSimon Westphahl proposed zuul/zuul master: Store runtime related system attributes together  https://review.opendev.org/c/zuul/zuul/+/80174908:33
opendevreviewSimon Westphahl proposed zuul/zuul master: Cache system config in Zookeeper  https://review.opendev.org/c/zuul/zuul/+/80185908:33
mhu1has anybody ever experienced mem leaks in nodepool-builder? I've just seen our process killed by OOMKiller, but not sure if it's just an outlier09:28
mhu1it's running on a machine with qemu-kvm, with 8GB RAM, so there should be plenty of room09:32
*** bhavikdbavishi1 is now known as bhavikdbavishi10:20
fungimhu1: our builders similarly have 8gb ram, but don't use much of it for anything besides cache: http://cacti.openstack.org/cacti/graph.php?action=view&local_graph_id=68028&rra_id=all (we also configure them with an additional 8gb swap just in case, which seems to go mostly untouched)11:15
opendevreviewSimon Westphahl proposed zuul/zuul master: Support cross scheduler config loading  https://review.opendev.org/c/zuul/zuul/+/77146311:16
fungigranted, our builders are running on xen domu guests at the moment, not kvm11:16
fungithough i doubt that difference would expose memory leaks11:18
*** dviroel|out is now known as dviroel11:18
*** jpena is now known as jpena|lunch11:20
*** jpena|lunch is now known as jpena11:23
*** jpena is now known as jpena|lunch11:30
opendevreviewSimon Westphahl proposed zuul/zuul master: Support cross scheduler config loading  https://review.opendev.org/c/zuul/zuul/+/77146312:13
mhu1fungi, ok thanks for the feedback. I'll keep an eye on it and see if it happens again12:26
*** jpena|lunch is now known as jpena12:33
opendevreviewSimon Westphahl proposed zuul/zuul master: Cache system config in Zookeeper  https://review.opendev.org/c/zuul/zuul/+/80185913:49
opendevreviewSimon Westphahl proposed zuul/zuul master: Support cross scheduler config loading  https://review.opendev.org/c/zuul/zuul/+/77146313:49
opendevreviewFelix Edel proposed zuul/zuul master: Add MergerApi  https://review.opendev.org/c/zuul/zuul/+/78768614:05
opendevreviewFelix Edel proposed zuul/zuul master: Fix test_gerrit.TestPolling.test_config_update  https://review.opendev.org/c/zuul/zuul/+/77302314:05
opendevreviewFelix Edel proposed zuul/zuul master: Execute merge jobs via ZooKeeper  https://review.opendev.org/c/zuul/zuul/+/80229914:05
opendevreviewFelix Edel proposed zuul/zuul master: Store merge params separate from request  https://review.opendev.org/c/zuul/zuul/+/80334714:05
opendevreviewFelix Edel proposed zuul/zuul master: Move merge request cleanup to Merger API  https://review.opendev.org/c/zuul/zuul/+/80334814:05
opendevreviewFelix Edel proposed zuul/zuul master: Clean up leaked merge request params  https://review.opendev.org/c/zuul/zuul/+/80334914:05
opendevreviewFelix Edel proposed zuul/zuul master: Don't use transactions when writing merge results  https://review.opendev.org/c/zuul/zuul/+/80335014:05
felixedelcorvus: I've uploaded a few follow-up changes to the Merger API that remove the transactions when writing the merge request params and deal with the cleanup of lost merge request params (similar to the Executor API). I've also rebased the whole stack on top of master (which already includes the relevant Executor API changes)14:09
Guest3154felixedel: great!14:10
felixedelIf you want you can squash them but I think it wouldn't harm to keep them separated. 14:11
felixedelAnd I think the last change https://review.opendev.org/c/zuul/zuul/+/803350/1 is failing the tests, but I haven't found the root cause yet. It seems to be related to the MergerResultEventFuture. Maybe I missed something when removing the transaction.14:13
felixedelIf you find time, feel free to continue with those patches, otherwise I will do so tomorrow14:14
Guest3154felixedel: sounds good14:17
Guest3154swest, tobiash, felixedel, clarkb: re clock skew: do you think 5m is too short?  i think we could easily lengthen it to 15m and be okay14:19
Guest3154these cleanups aren't time sensitive -- they just need to happen to clean up any potential leaks14:19
fungicorvus: your nick is showing up as "Guest3154" through the oftc bridge14:30
Guest3154fungi: ah.  it looks like there was an "event" earlier, and i think the bridge lost some traffic too.  :/14:31
fungineat14:32
swestcorvus: yeah I think I'd increase it a bit. Maybe even to 30min. I don't expect a lot of leaked event data nodes14:34
erbarrHi, I have a tenant where I'm loading the config project from my git, as an untrusted project I'm using opendev/ci-sandbox. In my config project, I have zuul.d with pipelines.yaml, projects.yaml, jobs.yaml. I believe I have defined a test job to just do an "ip a" in a check pipeline, however, I'm not seeing any check entry in the status page nor does the test job show up in the jobs list. I'm not seeing any errors pop up either, any 14:38
erbarridea what I could be doing wrong?14:38
Guest3154erbarr: check the scheduler log, it describes the choices zuul makes about enqueing changes and starting jobs14:39
erbarrGuest3154: thanks, location of this?14:39
*** Guest3154 is now known as notcorvus14:39
*** notcorvus is now known as corvus14:40
*** corvus is now known as Guest319714:40
Guest3197erbarr: are you the zuul system administrator?14:40
clarkbcorvus you flapped back to Guest after identifying14:41
clarkbmhu1: fungi: one thing to keep in mind is that the image build itself could be ooming14:41
erbarrGuest3197, yea, I'm installing the whole thing14:41
*** Guest3197 is now known as corvus14:42
*** corvus is now known as notcorvus14:42
*** notcorvus is now known as corvus14:42
*** corvus is now known as Guest320014:43
Guest3200erbarr: if you're running containers, then the container log; otherwise /var/log/zuul by default, or there's a zuul.conf option to supply your own log config14:43
erbarrI don't see a /var/log/zuul folder, I launched the scheduler, executor, and web, with the -f option though so I see the running process and I'm not seeing any errors. I had tried adding openstack/ironic as untrusted-project and that would lead to some errors with projects missing, which ended up in an ever growing list of run the scheduler and see what project is missing add it and try again, I was seeing no end to that. But anyways, I 14:50
erbarrwent simpler and no errors, just says that it loaded those files from my config project14:50
Guest3200erbarr: -f means run in foregroud and that disabled logging, so it'll go to stdout or stderr.  i'm assming you created a change.  you should see log entries with gerrit events in the scheduler logs.  check around those to see what zuul decided to do with it.14:52
Guest3200correction: not disabled logging, disabled writing to the log file.14:53
erbarrahh, ok, yea, I have a patch in ci-sandbox, are trigger events DEBUG log entries?14:54
*** jcapitao3 is now known as jcapitao14:55
Guest3200erbarr: if nothing at INFO helps, you may need to add "-d" to get debug.  zuul is a little less verbose about things it doesn't do at info level, generally focusing on things it does do.15:00
Guest3200and you're looking for a thing it's not doing :)15:00
erbarrlol, yea, I do see the event with the -d option, I'm trying to follow it15:01
Guest3200erbarr: there's an event id ([e: ...] in logs) that may help you trace it.  but note there still could be some relevant log entries without it15:02
erbarrok, thanks15:04
*** jpena is now known as jpena|off15:06
*** Guest3200 is now known as corvus15:17
*** corvus is now known as notcorvus15:17
*** notcorvus is now known as corvus15:17
corvusam i me?15:21
corvusyay i am15:21
fungiyou are!15:29
fungi(or a convincing imposter)15:30
opendevreviewBenjamin Schanzel proposed zuul/zuul master: Add tenant name on NodeRequests for Nodepool  https://review.opendev.org/c/zuul/zuul/+/78868015:30
*** rpittau is now known as rpittau|afk16:13
opendevreviewTristan Cacqueray proposed zuul/zuul master: [web] Update the TenantsPage to use React Hooks  https://review.opendev.org/c/zuul/zuul/+/80337516:14
tristanCcorvus: here is a quick draft for how functional component with react hook looks like: https://review.opendev.org/c/zuul/zuul/+/80337516:16
corvustristanC: thanks -- that's very helpful -- i have a followup Q that's better for email, so i'll reply there16:22
opendevreviewTristan Cacqueray proposed zuul/zuul master: [web] Update the TenantsPage to use React Hooks  https://review.opendev.org/c/zuul/zuul/+/80337516:27
tristanCcorvus: in PS2 I added back error management by dispatching an action to the existing error reducer16:28
*** marios|ruck is now known as marios|out16:30
*** ssbarnea[m] is now known as zbr16:31
corvustristanC: ++ replied in email16:36
tristanCgreat questions, let me add more details to the thread :-)16:41
opendevreviewTristan Cacqueray proposed zuul/zuul master: [web] Update the TenantsPage to use React Hooks  https://review.opendev.org/c/zuul/zuul/+/80337516:46
opendevreviewTristan Cacqueray proposed zuul/zuul master: [web] Update the TenantsPage to use React Hooks  https://review.opendev.org/c/zuul/zuul/+/80337517:27
*** dviroel is now known as dviroel|brb17:34
*** dviroel|brb is now known as dviroel18:42
opendevreviewlotorev vitaly proposed zuul/zuul master: DNM: testing job.required-projects inheritance  https://review.opendev.org/c/zuul/zuul/+/80339119:01
opendevreviewlotorev vitaly proposed zuul/zuul master: DNM: testing job.required-projects inheritance  https://review.opendev.org/c/zuul/zuul/+/80339119:08
*** dviroel is now known as dviroel|out20:54
opendevreviewlotorev vitaly proposed zuul/zuul master: doc: Specify job.required-projects is overriden  https://review.opendev.org/c/zuul/zuul/+/80326921:11
opendevreviewlotorev vitaly proposed zuul/zuul master: doc: Specify job.required-projects is extended by child jobs  https://review.opendev.org/c/zuul/zuul/+/80326921:12
opendevreviewJames E. Blair proposed zuul/zuul master: Clean up lost merge request results / data  https://review.opendev.org/c/zuul/zuul/+/80339821:35
clarkbcorvus: is ^ that stack next in the list of sos related changes?21:36
corvusclarkb: yes -- but i do need to squash 3 changes together21:42
clarkbok I'll wait then21:42
corvusfelixedel: (clarkb fyi) earlier you suggested keeping the original changes and the transaction-removal changes separate, however, because of the removal of transaction support from the underlying shard class, the originals don't work anymore, so i think we do need to squash them.  we also need to squash the change i just pushed up (803398), as it fixes a cleanup routine from the old code which doesn't work any more.21:43
corvusthat passes for me locally, so i'll squash them now, and then fixup a few things with the "execute merge jobs via zk" change too in just a minute21:44
corvuswell, few minutes -- this will involve some local test runs21:45
corvusfyi zuul folks -- we just accidentally had a really long debugging conversation in #opendev about stuck node requests after a zk restart.  i think we have identified the issue and have a fix planned.  i'll upload a change with an explanatory commit message in a bit23:24
opendevreviewJames E. Blair proposed zuul/zuul master: Add MergerApi  https://review.opendev.org/c/zuul/zuul/+/78768623:56
opendevreviewJames E. Blair proposed zuul/zuul master: Fix test_gerrit.TestPolling.test_config_update  https://review.opendev.org/c/zuul/zuul/+/77302323:56
opendevreviewJames E. Blair proposed zuul/zuul master: Execute merge jobs via ZooKeeper  https://review.opendev.org/c/zuul/zuul/+/80229923:56
corvusokay that's the squashed set23:57

Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!