[BUG] Actual size is showing unknown on volume detail page

[khushboo-rancher]

Describe the bug
Actual size is showing unknown on volume detail page for volume.

To Reproduce
Steps to reproduce the behavior:

1. In Rancher, while creating a deployment, select create PVC in storage tab.
2. Select Longhorn in storage class and fill out necessary fields.
3. Click create.
4. Go to Longhorn, check the volume created in volume details page.
5. The actual size is appearing as unknown

Expected behavior
The actual size should show the size properly.

Volume :

apiVersion: v1
items:
- apiVersion: longhorn.io/v1beta1
  kind: Volume
  metadata:
spec:
    Standby: false
    accessMode: rwo
    backingImage: ""
    baseImage: ""
    dataLocality: disabled
    disableFrontend: false
    diskSelector: null
    encrypted: false
    engineImage: longhornio/longhorn-engine:master-head
    fromBackup: ""
    frontend: blockdev
    lastAttachedBy: ""
    migratable: false
    migrationNodeID: ""
    nodeID: node_name
    nodeSelector: null
    numberOfReplicas: 3
    recurringJobs: null
    replicaAutoBalance: ignored
    revisionCounterDisabled: false
    size: "1073741824"
    staleReplicaTimeout: 2880
  status:
    actualSize: 51003392
    conditions:
      restore:
        lastProbeTime: ""
        lastTransitionTime: "2021-08-10T17:46:23Z"
        message: ""
        reason: ""
        status: "False"
        type: restore
      scheduled:

Environment:

• Longhorn version: Longhorn-master - c09ac49 08/10/2021
• Installation method (e.g. Rancher Catalog App/Helm/Kubectl): kubectl

[PhanLe1010]

Need to check the actualSize value returned by the API <LONGHORN-URL>/v1/volumes

[smallteeths]

The normal situation is that unKnow will not occur. The original pr is associated with an interface /v1/volumes/a1?action=snapshotList. If this interface request fails the actual size will have an unKnow status.

I think I can remove this condition now, since the real size of the volume is not associated with snapshotList.

The above bug is probably caused by the error returned by the /v1/volumes/a1?action=snapshotList interface.

[innobead]

The normal situation is that unKnow will not occur. The original pr is associated with an interface /v1/volumes/a1?action=snapshotList. If this interface request fails the actual size will have an unKnow status.

I think I can remove this condition now, since the real size of the volume is not associated with snapshotList.

The above bug is probably caused by the error returned by the /v1/volumes/a1?action=snapshotList interface.

cc @shuo-wu @jenting

[jenting]

I can't grep the commit id c09ac49 in longhorn-manager master branch.
But I found the f271ef470f706e58cbd0d30948e1f1189b028d8d would break the snapshotList API call.

Shuo is working on it.

[jenting]

After quickly sync up with Shuo, this case only happens if we access Longhorn GUI from the Rancher server.
If we access the Longhorn GUI by port-forward or NodePort, this issue won't happen.

[shuo-wu]

The root cause analysis and the fix are in #2869

[khushboo-rancher]

Closing after the validation of #2869