Thursday, 2022-06-23

*** rcastillo_ is now known as rcastill00:10
*** rlandy|bbl is now known as rlandy01:08
*** rlandy is now known as rlandy|out01:13
opendevreviewOpenStack Proposal Bot proposed openstack/project-config master: Normalize projects.yaml
*** pojadhav|out is now known as pojadhav|ruck02:38
fricklerone thing to note about the zuul ansible update is that the configuration says "ansible_version: 5", but e.g. in a job header we have "Ansible Version: 2.12.5"04:45
fricklerwe should either make that more explicit and say "Ansible Core Version" or try to find out the real ansible version04:46
ykarelfrickler, hi04:49
ykarelcan you please put autohold for puppet-glance-tripleo-standalone-1 and puppet-glance-tripleo-standalone-204:50
ykarelwant to debug an issue further, it's related to some node providers cpu architecture04:50
ykarela random issue, and not able to reproduce locally as not have those cpu archs04:51
*** ysandeep|out is now known as ysandeep04:53
ykarelianw, if you around please check ^04:59
ianwykarel: one sec05:00
ianwykarel: should be done ->
ykarelianw, Thanks05:06
opendevreviewMerged openstack/project-config master: Normalize projects.yaml
fricklerykarel: the jobs are already failing but stuck in post somehow. remind me of your ssh key and I'll give you access07:25
ykarelfrickler, THanks07:25
ykareli have access07:25
ykareladded keys in pre07:25
ykareli have added pause in post, that's why it stuck, will move soon07:26
fricklerykarel: ah, o.k., that pause confused me07:27
*** ysandeep is now known as ysandeep|afk07:30
*** jpena|off is now known as jpena07:38
ianwwe can do some cool things with zuul :) -> like screenshots of all our graphs as a CI output (wip, but basics working)
*** ysandeep|afk is now known as ysandeep09:20
*** deepcursion2 is now known as deepcursion09:53
*** rlandy_ is now known as rlandy10:34
*** rlandy_ is now known as rlandy10:49
*** rlandy_ is now known as rlandy11:08
*** dviroel|afk is now known as dviroel11:20
frickler show a held node 5 months old, but there is no matching autohold11:20
mariosfungi: clarkb: o/ hello sanity check please if you have a sec - that look right to you? (i.e. is that it?) 11:24
mariosfungi: clarkb: (sorry not sure why i hilighted line 42 there but i meant about moving the shared queue, "tripleo" in our case)11:25
fungimarios: yes, you've got it as a project-level attribute instead of pipeline-level in that change, so should be good11:26
mariosthanks fungi (lining up a whole bunch of them across our repos so wanted to be sure ;))11:27
fungithanks for working through it!11:28
mariosfungi:++ 11:28
*** pojadhav|ruck is now known as pojadhav|break11:32
*** pojadhav|break is now known as pojadhav|ruck11:44
*** rcastill is now known as rcastillo13:02
*** dasm|off is now known as dasm13:03
hrwthanks for doing all those zuul updates - now stream of console.log is printed with readable font ;D13:04
frickler redirects to , how can I drop that? exists and works fine13:06
fricklerhrw: are you talking about the live stream of a running job? because I don't see a difference there13:07
hrwfrickler: it was serif font before, now goes with sans-serif13:08
hrw'was' means '3 months ago' or sth13:09
fricklerhrw: I still see a courier-like serify font. which browser are you using?13:15
fungifrickler: you might be able to work around it with an exception in but i think the general idea is that if you're publishing to you'd use the standard openstack publishing jobs which put your content under /latest/13:28
fungifrickler: alternatively, if you want a landing page for the sig itself, you could do something similar to what i did for tact:
fricklerwell I intentionally used the promote-openstack-tox-docs-direct job to avoid the /latest/ subpath, because it doesn't make sense for a branchless information repo. so that is a standard job13:33
fricklerseems I would need to add it to like arch-design where I copied this from13:34
fungiahh, cool so there's already a precedent for other documents overriding it13:37
fungii had forgotten about www-generator.py13:38
*** ysandeep is now known as ysandeep|afk13:39
*** ysandeep|afk is now known as ysandeep13:47
hrwfrickler: firefox13:56
fricklerhrw: hmm, same here. then I have not idea what might be different for you13:57
hrwfrickler: maybe it finally just uses 'monospace' font instead of forcing courier ones13:57
ykarelfrickler, please delete the autoholds
ykareli am done with those, thanks14:10
ykarelit resulted into an ovs bug on some cpu models14:10
fricklerykarel: interesting, do you have a bug reference you can share? nodes are deleted now14:33
* ykarel leaves14:37
fricklerykarel: interesting read, thx14:43
Clark[m]Ya I've been sucked in by that. One thing I notice is that __get_cpuid_count does error checking but the oven code bypasses that error checking bu calling __cpuid_count which is an internal macro15:07
Clark[m]I've learned a whole lot about the cpuid instruction now. But if I had to guess that is the bug. Requesting info above what the CPU supports so getting junk back that masks as of the feature is present15:11
clarkbmarios: thank you for digging into the queue thing15:16
*** ysandeep is now known as ysandeep|out15:21
opendevreviewKendall Nelson proposed opendev/system-config master: Setup #openinfra-envirosig IRC Channel
fungidiablo_rojo_phone: ^ please also push the patch to add it to accesbot per my earlier review comment. 15:33
fungiwe need to merge the accessbot addition first15:33
diablo_rojo_phonefungi: on it!15:34
clarkbhrw: frickler: my first guess was maybe patternfly was selecting the serif font for us but their website says they use open sans. Maybe something the js terminal lib does ( a weird choice for a terminal lib)15:39
opendevreviewKendall Nelson proposed openstack/project-config master: Setup #openinfra-envirosig IRC Channel
opendevreviewKendall Nelson proposed opendev/system-config master: Setup #openinfra-envirosig IRC Channel
fungidiablo_rojo_phone: out of curiosity, is it an openstack sig or is there some new concept of foundation-level sigs?15:43
fungiand sorry to keep pinging you on your phone, it looks like your normal client isn't joined to this channel (but i see it in other channels)15:44
fricklerfungi: will accessbot also register the channel or would that need to be done manually first? I've always done the latter so far. -ChanServ( Channel #openinfra-envirosig is not registered with channel services.15:49
fungithe accessbot tests should error in check if the channel is not registered and configured for base access by accessbot15:52
fungiif it doesn't pass, odds are someone hasn't done
efoley #rhos-infrared16:06
diablo_rojo_phonefungi: sorry meeting hopping this morning, will respond better in a bit16:08
*** frenzy_friday is now known as frenzyfriday|PTO16:21
*** pojadhav|ruck is now known as pojadhav|afk16:26
*** jpena is now known as jpena|off17:02
diablo_rojo_phonefungi: Yeah I've been having issues lately with my IRC client not auto connecting to my channels so I have been doing it manually. Pinging me with this nick is totally fine. 17:04
diablo_rojo_phoneHopefully you kind of got the question answered (kinda) in the TC meeting?17:05
fungii tried to reflect that in my comment on 847364 yes17:34
diablo_rojo_phonefungi: thank you!18:05
TheJuliaHey OpenDev folks! Question, is rax the only provider which does pure qemu emulation, or are there others?18:07
clarkbTheJulia: I think you may need to clarify that a bit more. Are you asking if there are clouds which don't have nested virt enabled other than rax?18:08
TheJuliabasically yes18:08
clarkbok because to be clear we don't support nested virt at all its a use at your own risk due to its flakyness so while certain clouds may or may not have it present we don't really suggest people rely on it. For that reason I tend to not be fully up to date on who does and doesn't enable it.18:09
fungirackspace is xen-based so can't do nested kvm, but i believe we have some kvm-based providers which don't advertise nested kvm support either18:09
clarkbI know rax doesn't and the arm cloud don't (because arm doesn't do it at all)18:09
clarkbovh and vexxhost and inmotion do? I'm not sure about inaop18:10
TheJuliawe're encountering some jobs which are crawling under heavily cpu bound conditions where IO wouldn't necessarily be the issue. Traditionally we've only seen it on RAX in any way which is enough to cause a high job failure rate18:11
TheJuliait did help us find a bug though... but yeah18:12
fungithere are enough differences between providers that i wouldn't jump to the conclusion it's lack of nested kvm as a cause18:12
TheJuliayeah... I'm going to try and fix the race condition first and tune things a little bit more. See if we can cut down the amount of time a little bit more18:13
fungicould be rackspace regions have much more i/o contention, or are running on much older hardware, or... plenty of possibilitie18:13
TheJuliaor we're just unlucky on iweb 18:13
TheJuliano real way to know with the numbers we're seeing about digging in to every single job's logs in high detail18:13
clarkbcat /proc/cpuinfo | grep vmx returns nothing on iweb18:14
clarkbso ya I think rax, iweb, linaro, and osuosl all lack nested virt. The others have it but we don't really support it because it breaks often enough and is difficult to impossible for us to debug or fix18:17
TheJuliaack, thanks18:18
* TheJulia goes to tune the devstack plugin18:19
TheJuliais osuosl in the hostname?18:21
fungiyes, but that's arm64 based18:21
fungisame for linaro18:21
fungiso if you're not running arm jobs you won't see those18:21
*** rcastillo_ is now known as rcastillo18:28
*** dviroel is now known as dviroel|out21:14
opendevreviewSamuel Walladge proposed openstack/project-config master: Add Cinder Dell EMC PowerStore charm
opendevreviewGage Hugo proposed openstack/project-config master: Retire openstack-helm-deployments repo
opendevreviewGage Hugo proposed openstack/project-config master: Retire openstack-helm-deployments repo
*** dasm is now known as dasm|off22:07
*** rlandy is now known as rlandy|bbl22:30

Generated by 2.17.3 by Marius Gedminas - find it at!