*** bpradipt has joined #kata-dev | 04:59 | |
*** bpradipt has quit IRC | 06:03 | |
*** pcaruana has joined #kata-dev | 06:08 | |
*** bpradipt has joined #kata-dev | 06:20 | |
*** bpradipt has quit IRC | 06:21 | |
*** bpradipt has joined #kata-dev | 06:21 | |
*** amorenoz__ has joined #kata-dev | 06:41 | |
*** amorenoz has joined #kata-dev | 06:47 | |
*** amorenoz__ has quit IRC | 06:47 | |
*** sameo has joined #kata-dev | 07:10 | |
*** jodh has joined #kata-dev | 07:22 | |
*** bpradipt has quit IRC | 07:51 | |
*** davidgiluk has joined #kata-dev | 08:05 | |
*** bpradipt has joined #kata-dev | 08:18 | |
kata-irc-bot | <james.o.hunt> Hi @fupan - are you around? | 08:52 |
---|---|---|
kata-irc-bot | <fidencio> @bergwolf, hmmm. github is giving me a hard time to approve yours PRs | 09:18 |
kata-irc-bot | <fidencio> I'll get back to them later Today and hope that github will be behaving better by that time | 09:19 |
*** hashar has joined #kata-dev | 09:50 | |
fidencio | jodh: interesting, seems I don't have the power to have my "approvel" counting on ksm-throttler repo | 11:32 |
kata-irc-bot | <james.o.hunt> @fidencio - just added you to that repo. | 11:36 |
fidencio | jodh: thanks! | 11:36 |
*** pcaruana has quit IRC | 11:50 | |
*** pcaruana has joined #kata-dev | 12:02 | |
*** devimc has joined #kata-dev | 12:25 | |
*** vgoyal has joined #kata-dev | 12:38 | |
*** sameo has quit IRC | 12:49 | |
*** devimc has quit IRC | 12:55 | |
*** devimc has joined #kata-dev | 12:57 | |
*** sameo has joined #kata-dev | 13:04 | |
*** devimc has quit IRC | 13:48 | |
*** devimc has joined #kata-dev | 13:50 | |
kata-irc-bot | <fidencio> @salvador.fuentes, finally getting cri-o + kata (from 2.0-dev) installed on a f32 VM and I can try to reproduce the failures you've been facing | 13:54 |
*** crobinso has joined #kata-dev | 13:59 | |
kata-irc-bot | <salvador.fuentes> Hi @bergwolf, yes we are already using osbuilder inside kata-containers repo. What is missing is to make the switch of the packaging repo (to build kernel). Will make the change | 14:02 |
kata-irc-bot | <salvador.fuentes> Hi @fidencio, thanks for taking a look, I was talking with devimc and he says that the issue is that shimv2 does not support systemd cgroups and that is why those tests are failing | 14:03 |
kata-irc-bot | <eric.ernst> I’m curious what specifically about shimv2? | 14:05 |
*** dklyle has joined #kata-dev | 14:11 | |
kata-irc-bot | <bergwolf> cool, thanks! | 14:23 |
kata-irc-bot | <eric.ernst> You know @salvador.fuentes ? | 14:55 |
kata-irc-bot | <eric.ernst> Or just systemd at large? | 14:55 |
kata-irc-bot | <salvador.fuentes> not sure, devimc ^? | 14:55 |
devimc | sorry, nop | 14:56 |
kata-irc-bot | <salvador.fuentes> @eric.ernst opened this issue, in case you are interested https://github.com/kata-containers/kata-containers/issues/336 | 14:57 |
kata-irc-bot | <eric.ernst> Same config is fine on 1.x? | 14:58 |
kata-irc-bot | <salvador.fuentes> nop, same issue on both, but without using shimv2, crio tests with systemd slices work correctly | 15:02 |
fidencio | devimc: it's weird, I'm using shimv2 + cri-o + what I think to be systemd cgroups on OpenShift | 15:04 |
devimc | fidencio: sandbox_cgroup_only = true? | 15:06 |
fidencio | devimc: yes, but I'm using cgroups v1 (IIRC) | 15:06 |
devimc | weird | 15:07 |
*** hashar has quit IRC | 15:36 | |
*** sameo has quit IRC | 15:50 | |
*** hashar has joined #kata-dev | 15:52 | |
kata-irc-bot | <eric.ernst> @fidencio if you have it setup, what I would do is run 2 pods; one with runc on with Kata. | 15:55 |
kata-irc-bot | <eric.ernst> and compare the location within /sys/fs/cgroups | 15:55 |
kata-irc-bot | <eric.ernst> fidencio devimc ^ | 15:55 |
kata-irc-bot | <eric.ernst> I am working on adding cpuset support now and my next step would be to test on f32/systemd. | 15:56 |
*** ericernst has joined #kata-dev | 15:57 | |
kata-irc-bot | <fidencio> Right, I have it working and can do the test as soon as I finish my stuff here | 15:59 |
kata-irc-bot | <eric.ernst> thanks! | 16:00 |
*** sameo has joined #kata-dev | 16:09 | |
devimc | fidencio: can you monitor them using systemd-cgtop ? | 16:16 |
ericernst | i think systemd-cgtop and ls work regardless of which 'driver' you use? | 16:25 |
ericernst | or......... let me test one more time... | 16:26 |
fidencio | devimc: I'll check in a bit, when checking what ericernst asked :-) | 16:26 |
ericernst | yeah it does | 16:26 |
ericernst | i think it just scans sysfs, not just looking at slices. | 16:27 |
ericernst | on similar topic, devimc I just pushed updated PR for changing sandbox contraining for cpusets. | 16:27 |
ericernst | tested with docker update, etc; throwing into kube now to verify with static cpu manager | 16:28 |
ericernst | devimc: i don't think my PR changes the device handling, but I'll run a local test to verify. I'm wondering if we have any CI tests in place to check the whitelist today (or if we do *any* testing of sandbox cgroups only)? | 16:39 |
ericernst | have said that I see devices.list --> a *:* rwm | 16:41 |
*** ericernst has quit IRC | 16:44 | |
*** jodh has quit IRC | 16:52 | |
fidencio | @fuentess, okay, I can reproduce it here and I have to admit I'm ridiculously confused on how works in my 8.2 box | 16:59 |
fidencio | I'll install 8.2 and give it a try there as well | 16:59 |
*** hashar has quit IRC | 17:12 | |
*** bpradipt has quit IRC | 18:41 | |
*** ericernst has joined #kata-dev | 19:05 | |
*** ericernst has quit IRC | 19:08 | |
*** ericernst has joined #kata-dev | 19:08 | |
*** ericernst has quit IRC | 19:10 | |
*** ericernst has joined #kata-dev | 19:10 | |
*** ericernst has quit IRC | 19:14 | |
*** ericernst has joined #kata-dev | 19:14 | |
*** ericernst has quit IRC | 19:15 | |
*** davidgiluk has quit IRC | 19:16 | |
*** ericernst has joined #kata-dev | 19:16 | |
*** ericernst has joined #kata-dev | 19:17 | |
*** ericernst has quit IRC | 19:18 | |
*** ericernst has joined #kata-dev | 19:19 | |
*** hashar has joined #kata-dev | 19:25 | |
*** ericernst has quit IRC | 19:30 | |
kata-irc-bot | <eric.ernst> lame question on my end perhaps — what’s the trick to get all/more of the logs from v2-shim? | 20:49 |
kata-irc-bot | <eric.ernst> @jose.carlos.venegas.m @archana.m.shinde any hints? devimc ^ | 20:50 |
kata-irc-bot | <fidencio> using crio or containerd? | 20:50 |
kata-irc-bot | <eric.ernst> I am using containerd, but AFAIU it shouldn’t matter? | 20:51 |
kata-irc-bot | <eric.ernst> (unless, does containerd consume them all?) | 20:51 |
kata-irc-bot | <fidencio> https://github.com/kata-containers/documentation/blob/master/Developer-Guide.md#debug-logs-and-shimv2 | 20:52 |
kata-irc-bot | <fidencio> It does matter :slightly_smiling_face: | 20:52 |
kata-irc-bot | <eric.ernst> got it — just googled on docker pages and saw the same. | 20:52 |
kata-irc-bot | <eric.ernst> Thanks! | 20:52 |
kata-irc-bot | <fidencio> Unfortunately getting the logs is not as straightforward as it was with shimv1 | 20:52 |
kata-irc-bot | <fidencio> I'd love to have the containerd-shim-kata-v2 accepting the very same debug level as cri-o / containerd is running with | 20:53 |
kata-irc-bot | <eric.ernst> Yeah… | 20:54 |
kata-irc-bot | <eric.ernst> too much complications with that autodiscovery… | 20:54 |
*** devimc has quit IRC | 21:06 | |
kata-irc-bot | <jose.carlos.venegas.m> @eric.ernst I had the same problem last week and moved to classic kata to get logs :( | 21:18 |
*** hashar has quit IRC | 21:29 | |
kata-irc-bot | <eric.ernst> @jose.carlos.venegas.m with just enablign debug in the CRI impl, all is good. | 22:24 |
*** vgoyal has quit IRC | 22:57 | |
*** sameo has quit IRC | 23:34 | |
*** ericernst has joined #kata-dev | 23:58 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!