Paul Holzinger c0f4e2ce67 healthcheck: do not leak service on failed stop ...

We reset the failed unit to not leak it, however we did so before
stopping, this is wrong because when the stop fails we will again have a
failed unit. The correct thing is to reset after the stop because once
it is stopped it cannot create new errors.

I found this using the following reproducer and this is enough to fix
it:
```
while :; do
cid=$(podman run -d --name foo --health-cmd /home/podman/healthcheck \
	--health-startup-cmd /home/podman/healthcheck \
	quay.io/libpod/testimage:20241011 /home/podman/pause)
podman healthcheck run $cid
podman rm -fa
sleep 2
systemctl --user list-units --failed | grep $cid && break
done
```

Signed-off-by: Paul Holzinger <pholzing@redhat.com>

2024-10-25 15:27:05 +02:00

5.8 KiB

Raw Blame History

View Raw