Known issues

Storage system status stuck on 'Progressing'

Problem:

In some specific circumstances, after adding FlashSystem as external storage, some Flash storage systems might get stuck on 'Progressing' state due to a status sync delay on RedHat ODF operator.

Detected in version:

ODF 4.13 using ODF-FS 1.4.x

Problem verification:

On Openshift Console go to Storage -> Data Foundation -> storage systems. Some storage systems might be stuck forever with a status of: "Progressing" and never changes to "Available".

Storage-system-in-progressing-github3

Workaround:
  1. SSH into OCP cluster
  2. Switch to openshift-storage namespace by running:
    $ oc project openshift-storage
  3. Find odf-controller-manager pod by running:
    $ oc get pods | grep odf-operator-controller-manager
  4. Delete the pod found in the previous step by running:
    $ oc delete pod {odf-operator-controller-manager-*}
  5. The pod will be recreated automatically, verify pod creation by running:
    $ oc get pods | grep odf-operator-controller-manager
  6. Storage system status should change to 'Available' after a few minutes

ODF-FS Console pod failing due to OOMKilled failure

Problem:

The ODF-FS Console pod fails continuously in a CrashLoopBackOff. The ODF-FS operator reports the failure reason is due to OOMKilled (error 137).

Detected in version:

All ODF versions running ODF-FS

Problem verification:
  1. The ODF-FS Console pod failure reason can be extracted with the 'oc describe' command on the ODF-FS operator pod. Failure reason will be OOMKilled (error 137).
  2. Any attempt to delete the ODF-FS Console pod fails in the same way.
  3. The ODF-FS Console pod's log itself does not show any meaningful information as the pod cannot start.
Workaround:

  1. SSH into OCP cluster

  2. Switch to openshift-storage namespace by running:
    $ oc project openshift-storage

  3. Download and edit the ODF-FS operator subscription by running:
    $ oc edit subscription -n openshift-storage ibm-storage-odf-operator

  4. Add the following lines in the spec section. Please note the indentation, and that all should be lower case 

     config:
     resources:
         limits:
             cpu:     50m
             memory:  1000Mi
         requests:
             cpu:     50m
             memory:  1000Mi

  5. Save and close the file. The subscription will be automatically applied to the cluster, and the ODF-FS pods will be redeployed

  6. Confirm all the ODF-FS pods are functioning by running:
    $ oc get pods

Notes:

Changing the ODF-FS operator subscription will update the memory limit for all deployments in the subscription (ODF console, operator and sidecars).
However, changing the ODF-FS operator subscription will be preserved through upgrades, so it won't need to be changed again when upgrading to a future version.