02:04 <muradm> hi freedesktop

02:05 <muradm> i was looking XDG_ variables specifications and could not found any on specifications.freedesktop.org that will mention XDG_SESSION_ID

02:05 <muradm> is it unspecified?

05:28 <Adrinael> daniels, IGT is still bringing fdo-packet-m1xl-* to their knees fairly consistently: https://gitlab.freedesktop.org/drm/igt-gpu-tools/-/jobs/12606008

07:34 <bentiss> sigh, there are heaps of 'D' jobs on the 3 runners. I guess it's time for a reboot of them :(

07:36 <MrCooper> should maybe run IGT tests in VMs?

07:36 <bentiss> MrCooper: AFAICT, they are already running in VM: 407 root 20 0 151M 10960 4576 D 0.0 0.0 0:00.01 /usr/bin/qemu-aarch64 /builds/gfx-ci/igt-ci-tags/build/lib/tests/igt_simulation

07:36 <bentiss> but it doesn't terminate properly

07:38 <MrCooper> that's qemu-user, not a VM

07:38 <MrCooper> also, why does it run aarch64 tests via qemu on x86 instead of natively on aarch64 runners

07:39 <bentiss> they are running tests on all arches AFAICT: arm, aarch64, mips

07:39 <bentiss> htop

07:49 <bentiss> of course, gentle reboot is not working :(

08:00 <Adrinael> What's different between fdo-packet-m1xl-* and gst-htz-*? That jamming doesn't happen on the latter

08:02 <bentiss> Adrinael: not sure if they have qemu-user-mode there

08:05 <MrCooper> they have to (on the host), or Mesa's jobs would break

08:06 <MrCooper> maybe one or the other kind of runner has /dev/dri/* devices available?

08:06 <bentiss> then no ideas why the runners are behaving differently

08:07 <bentiss> on fdo we do have /dev/dri/card0 available, and it should be forwarded to the containers, yes

08:17 <bentiss> Adrinael: works for me too, less hassle to have to deal with it :)

08:17 <bentiss> cause the runners are pretty dead there, I have to force reboot them, I can not simply just ask them to reboot

08:18 <bentiss> which means more monitoring, etc...

08:19 <bentiss> FWIW, only asked to force reboot the third runner now

08:20 <bentiss> Adrinael: to target htz runners, you can ask the tag 'gstreamer'

08:20 <Adrinael> Thanks, was just wondering how the heck to do it

08:21 <bentiss> no worries

08:21 <bentiss> I'm glad they have set a tag :)

08:24 <bentiss> docker-free-space is still ruinning on m1xl-2, waiting for it to clean up space as we only had 100 GB free

08:28 <bentiss> and we are back to business

10:32 <aleksander> hey all! getting new CI pipeline issues today, not sure if someone else is experiencing them?

10:32 <aleksander> e.g. ERROR: Preparation failed: failed to pull image "registry.freedesktop.org/mobile-broadband/modemmanager/ubuntu/20.04:2021-07-28.2" with specified policies [always]: error pulling image configuration: unexpected EOF (manager.go:205:1s)

10:32 <aleksander> https://gitlab.freedesktop.org/mobile-broadband/ModemManager/-/jobs/12614945

10:32 <aleksander> any idea?

10:32 <bentiss> aleksander: the registry cache seems to be acting up, let me kick it

10:33 <aleksander> bentiss, thanks!

10:34 <bentiss> should be ready in a minute or so

10:36 <bentiss> ok, ready now

10:46 <__tim> hrm :)

10:47 <__tim> what kind of load are we talking about here (re. gst-htz-* runners)?

10:54 <Adrinael> __tim, https://gitlab.freedesktop.org/drm/igt-gpu-tools/-/jobs/12606013 this kind

11:14 <__tim> ah ok

11:27 <__tim> (should be no problem from the looks of it)

11:33 <Adrinael> great

12:21 <aleksander> bentiss, lovely, thanks!

19:55 <muradm> hi freedesktop

19:56 <muradm> where does XDG_SESSION_ID is specified?