AWS provider pod failing with "panic: runtime error: invalid memory address or nil pointer dereference" about secrets-store-csi-driver-provider-aws

[jbct]

Thank you for the bug report, @egarbi. We will take a look

[YuvalShAz]

Hi @egarbi I see that in your supplied secrets.yaml provider class you seem to be using an ssmparameter objectType and a full arn as the objectName. However, for parameter store objects you must use the name and not the full arn. From the README:

For SSM Parameter Store, this must be the Name of the parameter and can not be a full ARN.

I've reproduced your problem using the full arn:

MountVolume.SetUp failed for volume "grafana-db-creds" : rpc error: code = Unavailable desc = failed to mount secrets store objects for pod default/pod, err: rpc error: code = Unavailable desc = error reading from server: EOF

And I've confirmed that the container starts up and mounts as intended when using just the name.

An additional note: I can't see your full deployment specification file so it's impossible to know, but if you check the examples/ExampleDeployment.yaml you'll see that the

volumeMounts:
  - name: grafana-db-creds
    mountPath: /mnt/secrets
    readOnly: true
env:
  - name: GF_SECURITY_ADMIN_USER
    valueFrom:
      secretKeyRef:
        name: grafana-db-creds
        key: username
  - name: GF_SECURITY_ADMIN_PASSWORD
    valueFrom:
      secretKeyRef:
        name: grafana-db-creds
        key: password

block is included in the containers section and is therefore not at the same indentation as volumes which is what your deployment specification seems to be.