| *** pojadhav|out is now known as pojadhav|ruck | 04:04 | |
| frickler | lbragstad: interesting, I had proposed an even more radical approach at https://review.opendev.org/c/openstack/devstack/+/780417 , but I didn't get around to ironing out the details yet | 05:08 |
|---|---|---|
| frickler | maybe I can also get people interested in my stack to disable the keystone admin endpoint https://review.opendev.org/c/openstack/devstack/+/777345/ | 05:11 |
| frickler | and to end this morning's advertising: ianw, yoctozepto, gmann could you please have another look at enabling the bullseye platform job? https://review.opendev.org/c/openstack/devstack/+/789083 | 05:14 |
| frickler | actually it might be interesting to see whether those qemu memory usage issues would happen with kolla, too. | 05:15 |
| frickler | and I missed one, this fix should allow to uncap PyYAML in g-r: https://review.opendev.org/c/openstack/devstack/+/814271 | 05:18 |
| yoctozepto | frickler: in kolla we spin up too few instances to spot it probably ;-) | 06:16 |
| frickler | yoctozepto: yeah, not via failing tests, but looking at the actual memory consumption of qemu processes would give a hint. I'll try to do that locally | 06:17 |
| yoctozepto | frickler: ok, that sounds reasonable :-) | 06:17 |
| yoctozepto | ah, remember to switch kvm off | 06:17 |
| yoctozepto | that might make a difference here | 06:18 |
| yoctozepto | I believe with plain qemu they should behave the same | 06:18 |
| yoctozepto | but good you have some time to check | 06:19 |
| yoctozepto | anyhow, after merging on devstack side | 06:19 |
| yoctozepto | I think it would be nice to propose Debian as a testing runtime | 06:19 |
| frickler | oh, good point. actually I could test that with devstack first, build two instances locally and disable kvm on one | 06:19 |
| yoctozepto | if you propose it, I can vote on it :-) | 06:20 |
| yoctozepto | mhm, that makes sense as well | 06:20 |
| frickler | yoctozepto: yes, after seeing kolla's plans yesterday, that would sound like a good idea. if only I knew some TC member ... ;) | 06:20 |
| yoctozepto | if only! | 06:21 |
| ianw | frickler: will do. we should have it tested in dib gate/glean as well, to keep our images sane, i have changes up | 06:51 |
| *** jpena|off is now known as jpena | 07:29 | |
| yoctozepto | frickler: added debian to TC's PTG's agenda | 08:16 |
| frickler | yoctozepto: thx, I'll try to join that if my neutron things go fast enough. | 08:21 |
| yoctozepto | frickler: it's today in the first session | 08:24 |
| frickler | yoctozepto: today? the pad says thursday for me | 08:27 |
| yoctozepto | frickler: sorry, tomorrow* I am multitasking too heavily to notice what I write, need to calm down | 08:40 |
| yoctozepto | but the first session | 08:40 |
| opendevreview | Balazs Gibizer proposed openstack/tempest master: Remove race from wait_for_interface_detach waiter https://review.opendev.org/c/openstack/tempest/+/812472 | 09:24 |
| opendevreview | Balazs Gibizer proposed openstack/tempest master: Test Network QoS minimum packet rate scenarios https://review.opendev.org/c/openstack/tempest/+/806257 | 09:24 |
| frickler | slaweq: yoctozepto: do you happen to know where we get the ovn pkgs from for debian in CI? locally I get "Package 'ovn-central' has no installation candidate" etc. | 10:15 |
| frickler | or wait, do we still run debian with ovs? | 10:15 |
| yoctozepto | frickler: ovs | 10:15 |
| slaweq | frickler: I know only that it is packaged in Ubuntu | 10:15 |
| yoctozepto | we discussed this yesterday too | 10:15 |
| slaweq | I never run it on Debian | 10:15 |
| yoctozepto | packages are still coming to debian | 10:16 |
| slaweq | frickler: You can always configure devstack to install it from source | 10:16 |
| yoctozepto | if you have time, you can ask zigo on #debian-openstack about the status | 10:16 |
| frickler | are there upstream pkgs? how does kolla do it, also build from source? | 10:17 |
| yoctozepto | frickler: no ovn images for now, that's the issue | 10:38 |
| *** dviroel|rover|afk is now known as dviroel|rover | 11:18 | |
| *** jpena is now known as jpena|lunch | 11:25 | |
| zigo | yoctozepto: What about Debian?!? | 11:46 |
| zigo | Oh... | 11:47 |
| zigo | yoctozepto: frickler: Bullseye was released without OVN. If you need it, I can upload it to bullseye-backports... | 11:47 |
| zigo | slaweq: ^ | 11:47 |
| zigo | It's still on my todo to pacakge OVS 2.16, I'm late on this... :/ | 11:48 |
| frickler | zigo: yes, please do that. we would like to make bullseye a first class testing platform for this cycle | 11:53 |
| zigo | Oh, that'd be super nice. Will do this week then. | 11:53 |
| zigo | frickler: In the mean time, I can get it on non-official backports if you'd like (can be done within minutes...). | 11:54 |
| frickler | zigo: see https://review.opendev.org/c/openstack/devstack/+/789083 for the patch to enable debian as platform, currently still needs to use OVS | 11:54 |
| frickler | zigo: that'd be great too, I could use that for local testing | 11:54 |
| zigo | Wed. afteroon is meeting day for me, so probably I'll do that tomorrow. | 11:56 |
| lbragstad | frickler nice - reviewing your patch | 12:10 |
| lajoskatona | kpecmartin: Hi, regarding today's interop session with Neutron, in the etherpad (https://etherpad.opendev.org/p/neutron-yoga-ptg ) we have 15:00-15:30 UTC, is that still works for You? | 12:21 |
| lajoskatona | kopecmartin: which room will be used for this session? | 12:22 |
| opendevreview | Luigi Toscano proposed openstack/devstack-plugin-nfs master: Update the branch-specific jobs (+v, +w, +x, -s, -t) https://review.opendev.org/c/openstack/devstack-plugin-nfs/+/777078 | 12:23 |
| kopecmartin | lajoskatona: yep, i'm counting on it .. Neutron's room is fine, or should i book something else? | 12:23 |
| *** jpena|lunch is now known as jpena | 12:24 | |
| lbragstad | frickler i can abandon my patch in favor of yours | 12:30 |
| lbragstad | i just wasn't sure if we should wait to update the clouds.yaml until after the accounts exist, based on the original ordering | 12:31 |
| lajoskatona | kopecmartin: cool, it's kitsi by default, is that works for you? | 12:31 |
| *** pojadhav|ruck is now known as pojadhav|brb | 12:32 | |
| lajoskatona | kopecmartin: another thing which hit my eye (as last combing schedules) that nova has the same line: Interop-Nova - Wed (20th oct) 15:00 UTC - 15:30 UTC: (https://etherpad.opendev.org/p/nova-yoga-ptg l188.) | 12:32 |
| lajoskatona | yoctozepto: Hi, You had a common topic with dansmith regarding "gate health", see: http://lists.openstack.org/pipermail/openstack-discuss/2021-October/025223.html | 12:40 |
| kopecmartin | lajoskatona: you mean jitsi? yeah , it's fine .. oh, interesting, i thought the session with nova was at 16:00 .. anyway, that's not a problem as Arkady should attend the nova session, not me .. let me double check with him to make sure | 12:40 |
| lajoskatona | yoctozepto: do you have by chance a fresh list of used resources? This paste is quite old: http://paste.openstack.org/show/jD6kAP9tHk7PZr2nhv8h/ | 12:41 |
| lajoskatona | kopecmartin: yeah jitsi, sorry for my bad grammar..... Cool, than it's not like a session together with nova (we have a x-project with them anyway from 1400-1500 UTC :-)) | 12:42 |
| *** pojadhav|brb is now known as pojadhav|ruck | 12:43 | |
| frickler | yoctozepto: thanks for the hint about kvm, I can indeed reproduce the size-doubling locally simply by changing virt_type from kvm to qemu in nova.conf | 12:49 |
| frickler | next step is to check what happens with newer libvirt/qemu on hirsute/impish to rule out something debian-specific | 12:50 |
| gmann | frickler: +A on 789083, +2 on 814271. other two about keystone admin endpoint and export var I will check later | 12:59 |
| kopecmartin | gmann: hi, would it be possible to move the starter kits topic earlier (13:00 UTC) or later any time after 14:00 UTC? https://etherpad.opendev.org/p/tc-yoga-ptg | 13:28 |
| gmann | kopecmartin: let me check | 13:29 |
| gmann | kopecmartin: done, L41 https://etherpad.opendev.org/p/tc-yoga-ptg | 13:32 |
| kopecmartin | gmann: thank you | 13:32 |
| gmann | np! | 13:32 |
| opendevreview | Balazs Gibizer proposed openstack/tempest master: Remove race from wait_for_interface_detach waiter https://review.opendev.org/c/openstack/tempest/+/812472 | 13:36 |
| opendevreview | Balazs Gibizer proposed openstack/tempest master: Test Network QoS minimum packet rate scenarios https://review.opendev.org/c/openstack/tempest/+/806257 | 13:37 |
| frickler | yoctozepto: ok, same thing on hirsute, which has qemu-5.x similar to bullseye. wondering where to report that :-S | 13:53 |
| *** vishalmanchanda_ is now known as vishalmanchanda | 14:24 | |
| *** fabiooliveira_ is now known as fabiooliveira | 14:25 | |
| *** nahimsouza_ is now known as nahimsouza | 14:26 | |
| opendevreview | Merged openstack/devstack master: Enable running on Debian Bullseye https://review.opendev.org/c/openstack/devstack/+/789083 | 14:56 |
| lajoskatona | kopecmartin: we lost in nova cross-project, but finish soon.... | 15:01 |
| kopecmartin | lajoskatona: ack | 15:02 |
| lajoskatona | kopecmartin: we are back to https://meetpad.opendev.org/neutron-yoga-ptg | 15:05 |
| kopecmartin | lajoskatona: i'm there | 15:05 |
| yoctozepto | frickler: yeah, it seems to be new qemu | 15:44 |
| yoctozepto | frickler: centos has 5.1 | 15:45 |
| yoctozepto | could you check it? | 15:45 |
| yoctozepto | and re: kvm hint - you are very welcome; I tend to have many ideas but too little time to investigate by myself | 15:46 |
| yoctozepto | lajoskatona: no, I don't have anything fresher than that; I think infra helped with providing the number, dansmith to the rescue | 15:47 |
| lajoskatona | yoctozepto: thanks, dansmith forwarded me to clarkb, so check with him later | 15:50 |
| frickler | yoctozepto: I need to pick up some motivation to test on centos really. do you have a link for some cloud image handy? | 15:51 |
| yoctozepto | frickler: you mean: https://cloud.centos.org/centos/8-stream/x86_64/images/ ? | 15:55 |
| yoctozepto | lajoskatona: :-) | 15:56 |
| yoctozepto | frickler: just noticed qemu 5.1.0 comes from rdo so you have to install also centos-release-openstack package or something similar | 15:57 |
| yoctozepto | but I guess we have to report to qemu nonetheless | 15:57 |
| yoctozepto | it probably reserves the memory twice or something | 15:58 |
| yoctozepto | frickler: btw, I don't know if you saw it but it's linked from the whiteboard as well: https://docs.google.com/spreadsheets/d/1MS2KEhr5yMXEjD5ywU8peJomBNT0Va8bN8wy8hvodVY/edit#gid=0 | 16:06 |
| yoctozepto | switching to a single image would help tremendously with tracking the changes to versions | 16:06 |
| yoctozepto | it's great to debug some weird issues that happen from time to time | 16:06 |
| frickler | yoctozepto: oh, I hadn't seen that one yet, nice | 16:11 |
| *** pojadhav is now known as pojadhav|ruck | 16:12 | |
| *** pojadhav|ruck is now known as pojadhav|out | 16:12 | |
| frickler | fwiw I just did my first run on jammy, seems to deploy just fine | 16:13 |
| *** jpena is now known as jpena|off | 16:31 | |
| yoctozepto | you mean with kolla? | 16:49 |
| frickler | no, devstack, we are in -qa here, after all ;) | 19:56 |
| frickler | wow, 1.3G for cs8, what do they have in there, full desktop env for a cloud-image? | 20:04 |
| *** dviroel|rover is now known as dviroel|rover|afk | 22:53 | |
| *** dviroel|rover|afk is now known as dviroel|rover|out | 23:16 | |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!