Update values.yaml - #50
Conversation
Add - Example for SMTP with Username and Password - Fix Helm-Deploy-Warnings in case of duplicated Ports - Add PVC to store Grafana Userchanges and keep it after Helm-Delete
There was a problem hiding this comment.
I've tested the changes myself and suprisingly eveything still works. I dug a bit deeper and I think there's no point in updating this here, as this is the default, unfortunately expected behavior.
We can keep the other changes though. Thank you very much for your pull request!
| # Persists a PVC and keep it by Helm-Delete so you can Update and your local Changes should be available | ||
| persistence: | ||
| type: pvc | ||
| enabled: true |
There was a problem hiding this comment.
This should be false please. We don't want to create PVCs per default. If a user wants to persist their grafana dashboards they should be doing it declaratively and not by imperatively saving everything in the dashboard itself and then relying on a PV being there.
The size of the PVC is also way too large, if it's only for dashboards and grafana configuration.
Or ist there some other reason for this change?
| # Change Container-Port in Case of Helm Duplicate Port Warnings | ||
| - containerPort: 8081 |
There was a problem hiding this comment.
Thanks for working on this! I wanted to clarify what's happening with the port configuration to help explain why this
change works (and what it actually does) and why we just need to accept the warning for the time being.
What's Actually Happening
The duplicate port warning occurs because of how the upstream Grafana Helm chart works. When we set
service.targetPort: 8080 in the values.yaml, the Grafana chart automatically adds containerPort: 8080 to the Grafana container itself
(see line 1023 in the chart's _pod.tpl - https://github.com/grafana/helm-charts/blob/main/charts/grafana/templates/_pod.tpl). We then add the additional containerPort for the additional proxy container, which is exactly how the upstream chart does it (https://github.com/grafana/helm-charts/blob/main/charts/grafana/values.yaml#L388-L403).
Why Your Change "Works"
Changing the extraContainer port to 8081 removes the warning, but it's important to understand what's actually
changed:
- You removed the duplicate port warning (no more two containers claiming port 8080)
- The nginx proxy still listens on port 8080 (defined in the nginx.conf ConfigMap)
- The Service still targets port 8080 (defined in service.targetPort )
The containerPort field is mostly cosmetic/documentary - Kubernetes doesn't enforce it. So now:
- The Grafana container claims port 8081 (but doesn't actually use it)
- The nginx-proxy container claims port 8081 (but actually listens on 8080)
- The Service routes to port 8080 (where nginx is really listening)
Everything works because the Service routing is what matters, not the port declarations.
The Root Cause
This is a limitation of the upstream Grafana chart - it assumes the Service targets Grafana directly, not a sidecar
proxy. When using the proxy pattern, there's no clean way to avoid this mismatch without modifying the nginx.conf and service.targetPort together.
2 participants
Add