Repository navigation
Description
Describe your Issue
After updating the chart from 8.5.0 to 8.5.1 the nextcloud pod won't successfully start.
It's entering a Crashloopbackoff-state.
Logs and Errors
The pods log itself states "Upgrading nextcloud from 32.0.0.13 ...."
kubectl describe pod nextcloud-6845ccd787-mrlbz returns this:
Normal Scheduled 8m44s default-scheduler Successfully assigned default/nextcloud-6845ccd787-mrlbz to bastet
Normal Pulled 8m44s kubelet Container image "quay.io/k8tz/k8tz:0.18.0" already present on machine
Normal Created 8m44s kubelet Created container: k8tz
Normal Started 8m44s kubelet Started container k8tz
Normal Pulling 8m43s kubelet Pulling image "nextcloud:32.0.1-apache"
Normal Pulled 7m20s kubelet Successfully pulled image "nextcloud:32.0.1-apache" in 1m22.692s (1m22.692s including waiting). Image size: 523993624 bytes.
Warning Unhealthy 4m34s (x9 over 7m4s) kubelet Liveness probe failed: Get "http://10.42.1.195:80/status.php": dial tcp 10.42.1.195:80: connect: connection refused
Normal Started 3m4s (x5 over 7m20s) kubelet Started container nextcloud
Normal Killing 2m34s (x5 over 6m44s) kubelet Container nextcloud failed liveness probe, will be restarted
Normal Created 2m4s (x6 over 7m20s) kubelet Created container: nextcloud
Warning Unhealthy 2m4s (x32 over 7m8s) kubelet Readiness probe failed: Get "http://10.42.1.195:80/status.php": dial tcp 10.42.1.195:80: connect: connection refused
Normal Pulled 2m4s (x5 over 6m13s) kubelet Container image "nextcloud:32.0.1-apache" already present on machine
I guess the upgrading isn't fast enough, so the liveness probe fails and therefore the pod gets restarted, which results in a loop. (hence Crashloopbackoff)
Describe your Environment
-
Kubernetes distribution: k3s
-
Helm Version (or App that manages helm): version.BuildInfo{Version:"v3.19.0", GitCommit:"3d8990f0836691f0229297773f3524598f46bda6", GitTreeState:"clean", GoVersion:"go1.24.6"}
-
Helm Chart Version: 8.5.1
-
values.yaml:
# paste your values.yaml (anonymize any sensitive data)Additional context, if any
Not posting my values.yaml or other infos, since I think they're unnecessary.
Reverting back to 8.5.0 makes the nextcloud pod start up again.
Thx for your help & work.
If you need more info, please ask and I'll try provide it.
Activity
mreiche commentedon Dec 13, 2025
Check the output of
http://10.42.1.195:80/status.php, this may tell you the actual issue.Teonyr commentedon Dec 15, 2025
Since the issue is almost two months old, I just solved it in the meantime by temporarily disabling the probes for the update and enabling them again afterwards.
But thanks for your reply.
Maybe it'll help others.
artichoked1 commentedon Mar 2, 2026
This issue has always been a problem during major upgrades and never gets resolved. The nextcloud container returns a 500 during upgrade, so the readiness probe kills it halfway though.
I had better luck when using the fpm image - the fpm container is kept alive during upgrade and I just let the nginx container crashloop until the upgrade was finished.
artichoked1 commentedon Mar 2, 2026
It just occurred to me that a helm hook running a pre upgrade k8s
Jobcould be a clean fix to this? I'd be happy to submit a PR if I get time.I believe this issue to be quite important because if the probes kill the app mid-upgrade, it could cause a permanently botched upgrade.