Friday, 2020-09-18

*** armax has quit IRC00:17
*** armax has joined #openstack-infra00:21
*** ricolin has quit IRC00:22
*** hamalq has quit IRC00:30
*** armax has quit IRC00:34
*** rfolco|ruck has joined #openstack-infra00:38
*** rfolco|ruck has quit IRC00:42
*** gyee has quit IRC01:03
openstackgerritMerged openstack/project-config master: Remove noop jobs for python-adjutantclient  https://review.opendev.org/75099001:05
*** armax has joined #openstack-infra01:12
*** zbr7 has joined #openstack-infra01:16
*** Diabelko has joined #openstack-infra01:19
*** zbr has quit IRC01:19
*** rchanter has quit IRC01:19
*** systemd has quit IRC01:19
*** zbr7 is now known as zbr01:19
*** rchanter has joined #openstack-infra01:19
*** zzzeek has quit IRC01:21
*** zzzeek has joined #openstack-infra01:22
*** rfolco|ruck has joined #openstack-infra01:36
*** hamalq has joined #openstack-infra01:37
*** rfolco|ruck has quit IRC01:41
*** hamalq_ has joined #openstack-infra01:41
*** hamalq has quit IRC01:44
*** rcernin has quit IRC02:29
*** lbragstad has quit IRC02:31
*** zzzeek has quit IRC02:54
*** zzzeek has joined #openstack-infra02:54
*** ramishra has joined #openstack-infra02:58
*** zzzeek has quit IRC03:01
*** zzzeek has joined #openstack-infra03:04
*** rcernin has joined #openstack-infra03:15
*** dave-mccowan has quit IRC03:16
*** hamalq_ has quit IRC03:18
*** ricolin has joined #openstack-infra03:23
*** psachin has joined #openstack-infra03:35
*** psachin has quit IRC03:36
*** psachin has joined #openstack-infra03:37
*** ricolin_ has joined #openstack-infra03:42
*** ykarel|away has joined #openstack-infra04:01
*** ianychoi has quit IRC04:29
*** evrardjp has quit IRC04:33
*** evrardjp has joined #openstack-infra04:33
*** ykarel|away is now known as ykarel05:00
openstackgerritMerged openstack/project-config master: Remove openstack-python-jobs from GBP templates  https://review.opendev.org/75257505:05
*** zzzeek has quit IRC05:19
*** zzzeek has joined #openstack-infra05:19
*** xek has joined #openstack-infra05:26
*** vishalmanchanda has joined #openstack-infra05:33
*** lmiccini has joined #openstack-infra05:38
*** hamalq has joined #openstack-infra05:44
*** matt_kosut has joined #openstack-infra05:49
*** xek has quit IRC06:04
*** ysandeep|away is now known as ysandeep06:07
*** Tengu has quit IRC06:07
*** ralonsoh has joined #openstack-infra06:08
*** slaweq has joined #openstack-infra06:15
*** eolivare has joined #openstack-infra06:21
*** auristor has quit IRC06:21
*** auristor has joined #openstack-infra06:28
*** hamalq has quit IRC06:29
*** hamalq has joined #openstack-infra06:32
*** ramishra has quit IRC06:41
*** dklyle has quit IRC06:41
*** hamalq has quit IRC06:44
*** zbr5 has joined #openstack-infra06:45
*** amorin has quit IRC06:46
*** zbr5 has quit IRC06:49
*** zbr has quit IRC06:49
*** hamalq has joined #openstack-infra06:52
*** samueldmq has quit IRC06:52
*** yonglihe has quit IRC06:53
*** portdirect has quit IRC06:53
*** seongsoocho has quit IRC06:53
*** coreycb has quit IRC06:53
*** guilhermesp has quit IRC06:53
*** gagehugo has quit IRC06:53
*** dougwig has quit IRC06:53
*** cyberpear has quit IRC06:53
*** mwhahaha has quit IRC06:53
*** hogepodge has quit IRC06:53
*** johnsom has quit IRC06:53
*** vdrok has quit IRC06:53
*** rpittau|afk has quit IRC06:53
*** nicolasbock has quit IRC06:53
*** srwilkers has quit IRC06:53
*** donnyd has quit IRC06:53
*** csatari has quit IRC06:53
*** rm_work has quit IRC06:53
*** masayukig has quit IRC06:53
*** TheJulia has quit IRC06:53
*** ildikov has quit IRC06:53
*** mattmceuen has quit IRC06:53
*** samueldmq has joined #openstack-infra06:55
*** gagehugo has joined #openstack-infra06:55
*** nicolasbock has joined #openstack-infra06:55
*** zbr has joined #openstack-infra06:55
*** mwhahaha has joined #openstack-infra06:55
*** dougwig has joined #openstack-infra06:55
*** mattmceuen has joined #openstack-infra06:55
*** seongsoocho has joined #openstack-infra06:55
*** vdrok has joined #openstack-infra06:56
*** yonglihe has joined #openstack-infra06:56
*** ildikov has joined #openstack-infra06:56
*** portdirect has joined #openstack-infra06:56
*** coreycb has joined #openstack-infra06:56
*** guilhermesp has joined #openstack-infra06:56
*** masayukig has joined #openstack-infra06:56
*** cyberpear has joined #openstack-infra06:56
*** srwilkers has joined #openstack-infra06:56
*** hashar has joined #openstack-infra06:56
*** rpittau|afk has joined #openstack-infra06:56
*** donnyd has joined #openstack-infra06:57
*** csatari has joined #openstack-infra06:57
*** TheJulia has joined #openstack-infra06:57
*** johnsom has joined #openstack-infra06:57
*** hogepodge has joined #openstack-infra06:59
*** debian has joined #openstack-infra06:59
*** debian is now known as Guest6871106:59
*** Guest68711 has quit IRC07:00
*** debian1 has joined #openstack-infra07:00
*** andrewbonney has joined #openstack-infra07:01
*** debian1 has quit IRC07:02
*** amorin has joined #openstack-infra07:02
*** rchanter has quit IRC07:03
*** hamalq has quit IRC07:05
*** rm_work has joined #openstack-infra07:08
*** jcapitao has joined #openstack-infra07:10
*** derekh has joined #openstack-infra07:20
*** tosky has joined #openstack-infra07:25
*** gfidente|afk is now known as gfidente07:41
*** ociuhandu has joined #openstack-infra07:45
*** jpena|off is now known as jpena07:49
*** ykarel_ has joined #openstack-infra08:02
*** ramishra has joined #openstack-infra08:04
*** ykarel has quit IRC08:05
*** amorin has quit IRC08:08
*** ykarel_ is now known as ykarel08:10
*** lucasagomes has joined #openstack-infra08:14
*** ricolin_ has quit IRC08:22
*** dtantsur|afk is now known as dtantsur08:27
*** amorin has joined #openstack-infra08:51
*** mordred has quit IRC08:52
*** wolsen has quit IRC08:53
*** derekh has quit IRC08:55
*** mordred has joined #openstack-infra09:18
*** mordred has quit IRC09:19
*** derekh has joined #openstack-infra09:26
*** mordred has joined #openstack-infra09:26
*** vishalmanchanda has quit IRC09:43
*** Diabelko is now known as systemd09:53
*** vishalmanchanda has joined #openstack-infra09:57
*** wolsen has joined #openstack-infra10:08
*** psachin has quit IRC10:09
*** rcernin has quit IRC10:09
*** psachin has joined #openstack-infra10:10
*** jcapitao is now known as jcapitao_lunch10:16
*** psachin has quit IRC10:16
*** psachin has joined #openstack-infra10:17
*** xek has joined #openstack-infra10:33
*** rcernin has joined #openstack-infra10:44
*** hashar is now known as hasharAway10:47
*** andrewbonney has quit IRC10:51
*** rcernin has quit IRC10:54
*** ysandeep is now known as ysandeep|afk11:01
*** xek has quit IRC11:05
*** rcernin has joined #openstack-infra11:18
*** rcernin has quit IRC11:23
*** jpena is now known as jpena|lunch11:39
*** ysandeep|afk is now known as ysandeep11:42
*** hasharAway is now known as hashar11:51
*** rcernin has joined #openstack-infra11:54
*** jcapitao_lunch is now known as jcapitao11:59
*** rcernin has quit IRC11:59
*** ricolin_ has joined #openstack-infra12:02
*** rfolco|ruck has joined #openstack-infra12:07
*** rlandy has joined #openstack-infra12:07
*** andrewbonney has joined #openstack-infra12:36
*** jpena|lunch is now known as jpena12:37
*** Goneri has joined #openstack-infra12:42
*** dave-mccowan has joined #openstack-infra13:04
*** owalsh has quit IRC13:11
*** ysandeep is now known as ysandeep|session13:16
*** tosky_ has joined #openstack-infra13:18
*** tosky is now known as Guest8304713:18
*** tosky_ is now known as tosky13:18
*** xek has joined #openstack-infra13:36
dulekinfra-root: The frozen VM at 198.72.124.155 can be destroyed now, thanks a lot for help!13:49
fungidulek: were you able to identify the cause of your problem? curious what it was13:53
dulekfungi: Well, that's complicated. So apparently we were missing some local.conf option for OVN to work, yet when synchronizing them to match what's on Neutron gates I removed one that's required for Kuryr to work.13:55
*** rcernin has joined #openstack-infra13:55
dulekSo changes were fixing it but I broke another thing.13:56
dulekAnyway - I identified it and it seems to work now.13:56
fungioh, excellent13:56
fungiand thanks for the heads up, i'll release the node so it can be destroyed13:56
*** artom has joined #openstack-infra13:56
dulekfungi: But if we're talking already… So the OVN thing is just one of many stacked issues we face now. One is related to oom-killer that started to kill processes during our tests.13:57
dulekfungi: We're using Octavia and Amphora, so the mem footprint is very high, but this started to happen way more frequently last Friday.13:57
dulekDo you know if anything changed that could cause it?13:58
fungiinteresting, do you have an example build result from shortly before the problem started, for comparison?13:58
fungia number of jobs for master/victoria have been getting switched from ubuntu 18.04 to 20.04 in preparation for the release13:58
dulekWe were pretty close to the limit anyway, so it's possible something just started using 200 MB RAM more and we're toasted.13:59
fungipossible that happened in the timeframe you're seeing13:59
dulekHm, interesting, let me find you two results.13:59
fungiare you running dstat in your jobs? might be able to compare the dstat logs between them13:59
dulekfungi: We do, but dstat's has a bug so it shows one of our services is using TB of RAM. ;)14:00
*** rcernin has quit IRC14:00
fungioof, yeah that sounds wrong :/14:00
dulekSo this is fresh failed run: https://storage.bhs.cloud.ovh.net/v1/AUTH_dcaab5e32b234d56b626f72581e3644c/zuul_opendev_logs_ad9/752233/3/check/kuryr-kubernetes-tempest/ad96334/14:00
dulekIf I read it correctly it's still bionic.14:01
dulekAnd that's a successful run from a while ago: https://zuul.opendev.org/t/openstack/build/7bd9bf946e774326b099e97852f0c9f814:01
fungiyes, that looks like bionic to me14:01
fungi(the failed one)14:01
fungiso it's not the platform switch at least14:02
dulekNot like those help to debug, but they're pretty cool - mem stats from a run without the issue and with the issue:14:04
dulekhttps://i.imgur.com/jzKe5PG.png and https://i.imgur.com/nqojzRK.png14:04
fungidulek: you probably know without me having to dig in the setup logs, but are you not enabling swap? it's possible that could help in this case14:05
fungiat least enough to get a better picture of what's started using more ram14:06
dulekfungi: Uhm, honestly I didn't knew it's possible.14:06
dulekSo we probably don't.14:06
dulekBut that could help for sure, swapping out Amphora's should be fairly okay.14:06
*** xek has quit IRC14:09
fungidulek: looking at the failed run, it did create a 1gb swapfile14:10
fungi(the configure-swap role tasks)14:11
dulek"Setting up swapspace version 1, size = 8 GiB"14:12
dulekIsn't it 8 actually?14:12
*** hashar has quit IRC14:12
*** zxiiro has joined #openstack-infra14:12
*** hashar has joined #openstack-infra14:12
dulekfungi: OH DAMN IT.14:16
dulekhttps://opendev.org/openstack/openstack-zuul-jobs/commit/45f555fdf036de786b5988213b458b3b12dcef7414:16
dulekExactly this.14:16
dulekfungi: How can I overwrite that in our jobs?14:17
*** ysandeep|session is now known as ysandeep14:19
*** ysandeep is now known as ysandeep|away14:21
fungiaha! i thought for a moment to suggest the fallocate->dd switch, but ruled it out forgetting that it also reduced the swap size significantly14:25
fungidulek: i think configure_swap_size is probably plumbed through the devstack parent jobs... looking for an example now14:26
dulekfungi: I think there's none, but shouldn't putting it on `vars` in job config help?14:27
fungimaybe... the only example codesearch turns up outside the role itself is here: https://opendev.org/openstack/tripleo-ci/src/branch/master/roles/prepare-node/tasks/main.yaml#L314:27
fungithat's passing it directly to the role14:27
fungibut i think ansible will inherit the vars like you're suggesting14:28
dulekI'll check it. Thanks a lot for help, we're struggling with that for the whole week.14:28
fungieasy enough to see if you propose it14:28
fungiit should get applied speculatively, so just look at the resulting run and check the configure-swap output14:29
*** owalsh has joined #openstack-infra14:29
dulekYup!14:29
fungithough you may have to recheck a few times to hit a node with no ephemeral disk14:29
fungioh, nevermind, it uses that value when sizing swap partitions too14:30
fungiso yeah, you ought to be able to tell regardless of where it runs14:30
*** dmellado has quit IRC14:40
*** dmellado_ has joined #openstack-infra14:40
*** dmellado_ is now known as dmellado14:41
*** __ministry1 has joined #openstack-infra14:42
dulekLooks good: "8589934592 bytes (8.6 GB, 8.0 GiB) copied, 56.9489 s, 151 MB/s" :)14:43
dulekYay, what a way to end the work week!14:43
fungidulek: awesome!14:43
fungihard to beat a one-liner fix ;)14:43
fungidulek: if you have a few minutes, that may also be worth posting to openstack-discuss in case any other projects are struggling with the same symptoms14:44
dulekfungi: Yeah, I was planning to do that.14:44
fungii definitely didn't remember that we reduced the swap default when switching away from fallocate14:45
clarkbits because dd is slow14:58
clarkbthat will add 5 minutes to all your jobs ir something14:58
clarkbunfortunately ext4 and faalocate decided to stop working for swap files :(14:58
dulekclarkb: Well, that's better than having broken jobs.14:58
dulekWe weren't able to merge anything since last Friday.14:59
*** ianychoi has joined #openstack-infra14:59
clarkbdulek: I would argue that needing more than 1GB swap is broken ;)14:59
*** ykarel is now known as ykarel|away14:59
dulekclarkb: Oh, I can just make all of my jobs multinode instead to have more RAM. ;)14:59
clarkbits really there to keep a job that is on the cusp from failing. Not to ensure jobs can get by deep into swap14:59
clarkbdulek: yes thats the idea acyually15:00
dulekIsn't swap cheaper in that case?15:00
*** nightmare_unreal has quit IRC15:00
dulekOkay, anyway it's only because we're running Octavia and that means Amphora VMs and each uses 1 GB.15:01
clarkbin isolation likely. But heavy useof swap appears to be amajor contributor to our noisy neoghbor problwm15:01
dulekWe'll be switching to use ovn-octavia more soon, which makes this irrelevant.15:01
clarkbyou eatup all the disk iops15:01
johnsomNo, your VMs are about 150MB per your mem tracker15:01
clarkbthen a bunch of jobs fail in weird ways15:01
*** dklyle has joined #openstack-infra15:02
dulekjohnsom: Hm. So what's chewing up all that RAM on our gates?15:02
clarkbthe human side then spends a lot of cycles trying to understand the weird behavior15:02
dulekAnd I know, kuryr-daemon appears to use 1 TB of RAM, but that's dstat's bug.15:02
clarkbdulek: dstat's csv log should tell you15:02
dulekclarkb: Got it. We'll prioritize switch to ovn-octavia and should be able to drop that swap thing soon.15:03
johnsomYeah, I downloaded the memtracker data and loaded it up in a spreadsheet. You had four load balancers, next was 500MB mysql and a fairly large etcd15:03
dulekAnd then all the Python services. Interesting.15:04
johnsomThen a bunch of kube stuff and the openstack parts15:04
dulekjohnsom: Thanks for taking a look, I appreciate it!15:06
johnsomSure, NP. Dropping that thread count will help save you some RAM too.15:07
*** smarcet has joined #openstack-infra15:09
*** xek has joined #openstack-infra15:16
*** xek has quit IRC15:17
*** smarcet has quit IRC15:24
*** lmiccini has quit IRC15:27
*** smarcet has joined #openstack-infra15:42
*** d34dh0r53 has quit IRC15:43
*** d34dh0r53 has joined #openstack-infra15:44
*** tkajinam has quit IRC15:44
*** ykarel|away has quit IRC15:49
*** ricolin_ has quit IRC15:50
*** lucasagomes has quit IRC15:54
*** rcernin has joined #openstack-infra15:56
*** smarcet has quit IRC16:02
*** rcernin has quit IRC16:03
*** jcapitao has quit IRC16:08
*** eolivare has quit IRC16:08
*** dtantsur is now known as dtantsur|afk16:10
*** smarcet has joined #openstack-infra16:12
*** hashar has quit IRC16:21
*** vishalmanchanda has quit IRC16:23
*** ociuhandu_ has joined #openstack-infra16:23
*** ociuhandu has quit IRC16:27
*** ociuhandu_ has quit IRC16:28
*** ociuhandu has joined #openstack-infra16:33
*** ociuhandu has quit IRC16:37
*** __ministry1 has quit IRC16:41
*** psachin has quit IRC16:49
*** Tengu has joined #openstack-infra16:54
*** jpena is now known as jpena|off16:56
*** d34dh0r53 has quit IRC16:58
*** d34dh0r53 has joined #openstack-infra17:00
*** derekh has quit IRC17:01
*** hamalq has joined #openstack-infra17:01
*** iurygregory has quit IRC17:10
*** andrewbonney has quit IRC17:11
*** gfidente has quit IRC17:12
*** ralonsoh has quit IRC17:14
*** harlowja has joined #openstack-infra17:26
*** gyee has joined #openstack-infra17:35
openstackgerritThomas Bachman proposed openstack/project-config master: Remove legacy gate jobs  https://review.opendev.org/75273517:46
*** d34dh0r53 has quit IRC18:07
openstackgerritClark Boylan proposed openstack/project-config master: Remove old nodepool builder configs  https://review.opendev.org/75274118:14
*** d34dh0r53 has joined #openstack-infra18:15
*** irclogbot_1 has quit IRC18:19
*** irclogbot_1 has joined #openstack-infra18:23
*** irclogbot_1 has quit IRC18:31
*** irclogbot_2 has joined #openstack-infra18:35
openstackgerritAndreas Jaeger proposed openstack/project-config master: Switch grafana nodepool to Fedora 32  https://review.opendev.org/75274518:42
*** d34dh0r53 has quit IRC19:02
*** d34dh0r53 has joined #openstack-infra19:07
*** zxiiro has quit IRC19:17
*** zxiiro has joined #openstack-infra19:38
*** slaweq has quit IRC19:41
*** zzzeek has quit IRC20:06
*** zzzeek has joined #openstack-infra20:42
*** owalsh has quit IRC20:46
*** zzzeek has quit IRC20:48
*** zzzeek has joined #openstack-infra20:54
*** zzzeek has quit IRC20:59
*** harlowja has quit IRC21:01
*** ociuhandu has joined #openstack-infra21:02
*** zzzeek has joined #openstack-infra21:03
*** zzzeek has quit IRC21:05
*** zzzeek has joined #openstack-infra21:06
*** ociuhandu has quit IRC21:06
*** zzzeek has quit IRC21:13
*** matt_kosut has quit IRC21:14
*** owalsh has joined #openstack-infra21:14
*** armax has quit IRC21:15
*** zzzeek has joined #openstack-infra21:16
*** rlandy has quit IRC21:18
*** zzzeek has quit IRC21:23
*** zzzeek has joined #openstack-infra21:24
*** rcernin has joined #openstack-infra21:27
*** smarcet has quit IRC21:33
*** rcernin has quit IRC21:35
*** rcernin has joined #openstack-infra21:36
*** Goneri has quit IRC21:39
*** rcernin has quit IRC21:42
*** rcernin has joined #openstack-infra22:00
*** smarcet has joined #openstack-infra22:01
*** smarcet has quit IRC22:05
*** dave-mccowan has quit IRC22:22
*** zzzeek has quit IRC22:25
*** zzzeek has joined #openstack-infra22:26
*** zzzeek has quit IRC22:30
*** zzzeek has joined #openstack-infra22:31
*** eharney has quit IRC22:35
*** rfolco|ruck has quit IRC22:56
*** hamalq has quit IRC23:02
*** EmilienM has quit IRC23:06
*** EmilienM has joined #openstack-infra23:06
*** rcernin has quit IRC23:06
*** zzzeek has quit IRC23:08
*** zzzeek has joined #openstack-infra23:11
*** zzzeek has quit IRC23:16
*** zzzeek has joined #openstack-infra23:17
*** armax has joined #openstack-infra23:27
*** tosky has quit IRC23:31
*** rcernin has joined #openstack-infra23:38
*** Goneri has joined #openstack-infra23:44
*** hamalq has joined #openstack-infra23:48
*** armax has quit IRC23:54

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