Open garlick opened 1 year ago
Update: After swapping the docker-run-systest.sh
script to use podman --systemd=always ...
the el8,system
builder does not skip this sharness test, however it generates the following errors from the first (and all subsequent tests)
Jun 14 15:42:44.891567 UTC sdexec.debug[0]: watch 4ed92c07-f7e8.service
Jun 14 15:42:44.891593 UTC sdexec.debug[0]: start 4ed92c07-f7e8.service
Jun 14 15:42:44.911176 UTC sdexec.debug[0]: 4ed92c07-f7e8.service: unknown.unknown
Jun 14 15:42:44.911257 UTC sdexec.debug[0]: 4ed92c07-f7e8.service: activating.start
Jun 14 15:42:44.911976 UTC sdexec.debug[0]: 4ed92c07-f7e8.service: active.running
cat: /sys/fs/cgroup//user.slice/user-1001.slice/user@1001.service/4ed92c07-f7e8.service/memory.high: No such file or directory
Jun 14 15:42:44.917370 UTC sdexec.debug[0]: 4ed92c07-f7e8.service: active.running
Jun 14 15:42:44.917435 UTC sdexec.debug[0]: 4ed92c07-f7e8.service: failed.failed
Jun 14 15:42:44.917441 UTC sdexec.debug[0]: reset-failed 4ed92c07-f7e8.service
Jun 14 15:42:44.918027 UTC sdexec.debug[0]: 4ed92c07-f7e8.service: inactive.dead
Jun 14 15:42:44.918061 UTC sdexec.debug[0]: unwatch 4ed92c07-f7e8.service
0: Exit 1
This implies all the skip_all checks are passing, but the actual memory values aren't showing up in the service cgroup in the Github CI environement.
(Capturing this here in case anyone ever wants to dig into it)
Problem: #5359 adds a test
t2410-sdexec-memlimit.t
that is only run by theel8,system
builder in CI, but it is currently skipped.