traefik-helm-chart

traefik-helm-chart copied to clipboard

Welcome!

• [X] Yes, I've searched similar issues on GitHub and didn't find any.
• [X] Yes, I've searched similar issues on the Traefik community forum and didn't find any.

What version of the Traefik's Helm Chart are you using?

traefik-10.15.0

What version of Traefik are you using?

2.6.1

What did you do?

When using the optional initContainer to fix the volume permissions bug:

initContainers:
  - name: volume-permissions
    image: busybox:1.35.0
    command: ["sh", "-c", "chmod -Rv 600 /data/*"]
    volumeMounts:
      - name: data
        mountPath: /data

See https://github.com/traefik/traefik-helm-chart/blob/master/traefik/values.yaml#L47

The command should be changed to chmod -Rv 600 /data/* || true as to not fail if the directory mountPath does not exist which is the case on install. Without || true the init container breaks with:

pi@kube-master: $ kubectl logs traefik-7d8bf8b9b-7bqq2 -c volume-permissions
chmod: /data/*: No such file or directory

What did you see instead?

Failed init container on fresh install.

chmod: /data/*: No such file or directory

What is your environment & configuration?

Helm: 3.8.1 Kubernetes: 1.23.3

Additional Information

No response

same here any solution now ?

You could ignore errors:

initContainers:
  - name: volume-permissions
    image: busybox:1.35.0
    command: ["sh", "-c", "chmod -Rv 600 /data/* || echo oops"]
    volumeMounts:
      - name: data
        mountPath: /data

You could create/update some dummy file:

initContainers:
  - name: volume-permissions
    image: busybox:1.35.0
    command: ["sh", "-c", "date >/data/last-init ; chmod -Rv 600 /data/*"]
    volumeMounts:
      - name: data
        mountPath: /data

You could first deploy Traefik without this initContainer, and then enable this option once files were written in there.

Or you could use ext4 filesystem for your volume, in which case this /data/* would find at least some lost+found directory.

This initContainer configuration you mention isn't a default. Merely a sample. Obviously it would not suit all use cases -- and may not be necessary at all: chowning files is typically something kubernetes could do for you, attaching volumes to containers while a pod is starting up.

The question really being: what are you trying to do here? Do you need to chown anything to begin with? On each start?! If this is really the case, I would instead investigate on why those permissions needs to be changed in the first place.

Hello @nodesocket,

Following #164, we updated the workaround with PR #658.

Does it works for you ?

@mloiseleur ummm I have my command as just:

  initContainers:
    - name: volume-permissions
      image: busybox:1.35.0
      command: ["sh", "-c", "chmod -Rv 600 /data/* || true"]
      volumeMounts:
        - name: data
          mountPath: /data

That seems to be working consistently for me.

So I understand that your issue is fixed. Cool !

Where does the initContainers need to be added?

this should be in your values file. See: https://github.com/traefik/traefik-helm-chart/blob/e3d2f6e32f4d2e5ddb5c9f365d3836787a7a8218/traefik/values.yaml#L45-L54