bug: victoria-logs-cluster vlstorage.existingClaim doesn't work

[sahilarora535]

Chart name and version
chart: victoria-logs-cluster
version: v0.10.0

Describe the bug
I am using victoria-logs-cluster with two vlstorage pods. There is a key in vlstorage.persistentVolume.existingClaim. When this key is set, the expectation is to use the supplied persistent volume claim instead of creating a new claim. When this is unset, the helm-chart creates 2 pvc, <pvc-name>-0 and <pvc-name>-1.

What I did was I created the persistent volume claim myself with the same name. When I set vlstorage.persistentVolume.existingClaim: "victoria-logs-cluster-vlstorage-pvc" then the vlstorage pods don't come up.

Events:                                                                                                                                                                                                          │
│   Type     Reason             Age                 From                Message                                                                                                                                    │
│   ----     ------             ----                ----                -------                                                                                                                                    │
│   Warning  FailedScheduling   35m                 default-scheduler   0/12 nodes are available: persistentvolumeclaim "victoria-logs-cluster-vlstorage-pvc" not found. preemption: 0/12 nodes are available: 12  │
│ Preemption is not helpful for scheduling.                                                                                                                                                                        │
│   Warning  FailedScheduling   20m (x3 over 30m)   default-scheduler   0/12 nodes are available: persistentvolumeclaim "victoria-logs-cluster-vlstorage-pvc" not found. preemption: 0/12 nodes are available: 12  │
│ Preemption is not helpful for scheduling.                                                                                        

This is happening because of the fact that I created two PVCs, which were:

victoria-logs-cluster-vlstorage-pvc-0
victoria-logs-cluster-vlstorage-pvc-1

And instead of following the statefulset indexing for the existing claim, it's using the claim name as is, which it's unable to find.

This looks like a gap in helm chart. It would be greatly helpful if this could be fixed.

Custom values
Please provide only custom values (excluding default ones):

  persistentVolume:
    # -- Create/use Persistent Volume Claim for vlstorage component. Empty dir if false. If true,  vlstorage will create/use a Persistent Volume Claim
    enabled: true
    name: vlstorage-volume

    # -- Array of access modes. Must match those of existing PV or dynamic provisioner. Details are [here](https://kubernetes.io/docs/concepts/storage/persistent-volumes/)
    accessModes:
      - ReadWriteOnce
    # -- Persistent volume annotations
    annotations: {}
    # -- Persistent volume labels
    labels: {}
    # -- Storage class name. Will be empty if not set
    storageClassName: ""
    # --  Existing Claim name. Requires vlstorage.persistentVolume.enabled: true. If defined, PVC must be created manually before volume will be bound
    existingClaim: "victoria-logs-cluster-vlstorage-pvc"

[AndrewChubatiuk]

hey @sahilarora535
have you tried using vlstorage.persistentVolume.name instead?

[AndrewChubatiuk]

problem is that StatefulSet expects existing claim to be in format <claim-name>-<statefulset-name>-<idx> and it should be referenced as persistentVolume.name: <claim-name>. also spec.volumeClaimTemplate.spec.resources.requests.storage should be set explicitly. in your case naming convention is <statefulset-name>-<idx>. you need to:

• set reclaim policy to Retain
• delete old claims
• update retained PVC claimRef that should reference new claims that match naming convention <claim-name>-<statefulset-name>-<idx>
• create these claims
• check using kubectl that they are bound

only after these steps you can set .Values.vlstorage.persistentVolume.name: <claim-name> and install cluster chart
also .Values.vlstorage.persistentVolume.size should match value in your PVCs