| slangasek | acheronuk[m]: it's still a bug for autopkgtests to rely on $HOME to be set somewhere useful in the environment; nevertheless I would certainly like to know why all of a sudden, autopkgtests are running in the armhf containers as user 'systemd-coredump'. | 04:26 |
|---|---|---|
| slangasek | acheronuk[m]: it looks like a preponderance of the failing tests were on a node I was manually fiddling with to try to debug snapcraft failures, and the lxd profile may not have been 100% restored, sorry. Testing a possible fix now. | 04:32 |
| slangasek | hmmm but here's another failure on a different node | 04:34 |
| slangasek | * adt-virt-lxc: Stop hardcoding the "ubuntu" suggested normal user; instead, | 04:48 |
| slangasek | call "getent passwd" to detect the first user with uid >= 500. Don't | 04:48 |
| slangasek | advertise suggested-normal-user if none exists. | 04:48 |
| * slangasek sighs | 04:48 | |
| slangasek | xnox: ^^ where does this systemd-coredump user come from? it appears to be in none of the systemd maintainer scripts and is not mentioned in the changelog | 04:56 |
| slangasek | xnox: answer: /usr/lib/sysusers.d/systemd.conf but why and how is it picking uids? | 04:59 |
| slangasek | is it valid for systemd-coredump to have uid 998? yes. Is it weird? also yes. | 05:00 |
| slangasek | anyway, I think I have the autopkgtest harness fixed; force-restarting | 05:04 |
| acheronuk | slangasek: I see. thank you | 06:20 |
| === M_hc[m] is now known as _hc | ||
| === enick_48 is now known as RAOF | ||
| -queuebot:#ubuntu-release- New binary: node-rollup [amd64] (cosmic-proposed/universe) [0.50.0-1build1] (no packageset) | 23:21 | |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!