Flatcar
Flatcar copied to clipboard
Unable to SSH after OOM kill of systemd user manager units
Description
We noticed on a Flatcar instance that two systemd user manager units have been OOM killed: user@500 and user@1000.
This is OK but for unknown reasons yet it prevents SSH access to the instance until the reboot of the instance.
Additional information
- I was able to run
ssh core@localhost, of course the authentication has failed but it shows that SSH agent was responding sshd.socketwas up and running andss -lnp | grep 22was returning a listening state
core@bld-fairlie ~ $ journalctl --boot 10e7cf52b3ac4fa7990b21d57a89f294 -u user@500
May 11 12:09:49 bld-fairlie systemd[1]: Starting [email protected] - User Manager for UID 500...
May 11 12:09:49 bld-fairlie (systemd)[2257]: pam_unix(systemd-user:session): session opened for user core(uid=500) by (uid=0)
May 11 12:09:49 bld-fairlie systemd[2257]: Queued start job for default target default.target.
May 11 12:09:49 bld-fairlie systemd[2257]: Created slice app.slice - User Application Slice.
May 11 12:09:49 bld-fairlie systemd[2257]: Reached target paths.target - Paths.
May 11 12:09:49 bld-fairlie systemd[2257]: Reached target timers.target - Timers.
May 11 12:09:49 bld-fairlie systemd[2257]: Starting dbus.socket - D-Bus User Message Bus Socket...
May 11 12:09:49 bld-fairlie systemd[2257]: Listening on dbus.socket - D-Bus User Message Bus Socket.
May 11 12:09:49 bld-fairlie systemd[2257]: Reached target sockets.target - Sockets.
May 11 12:09:49 bld-fairlie systemd[2257]: Reached target basic.target - Basic System.
May 11 12:09:49 bld-fairlie systemd[2257]: Reached target default.target - Main User Target.
May 11 12:09:49 bld-fairlie systemd[2257]: Startup finished in 213ms.
May 11 12:09:49 bld-fairlie systemd[1]: Started [email protected] - User Manager for UID 500.
May 15 00:15:20 bld-fairlie systemd[1]: [email protected]: A process of this unit has been killed by the OOM killer.
May 15 00:15:20 bld-fairlie systemd[1]: [email protected]: Main process exited, code=killed, status=9/KILL
May 15 00:15:20 bld-fairlie systemd[1]: [email protected]: Failed with result 'signal'.
May 15 00:15:20 bld-fairlie systemd[1]: [email protected]: Consumed 32.124s CPU time, 12.1M memory peak.