Post-installation checks

After you deploy Fusion Data Foundation with Multus, verify that network assignments, storage access, and cluster behavior work as expected.

In addition to the Fusion Data Foundation health checking, ensure that Multus is configured as expected by performing the following:

• Using the ODF toolbox or odf-cli tool, run ceph osd dump to verify that OSDs have an IP in the cluster network.

  In the output, ensure that OSD daemons have an IP in the public network (if used). OSDs should additionally have an IP in the cluster network, if the Multus cluster network was used. If it was left empty, OSDs should only list public network IPs. If a filesystem has been deployed, ensure that MDS daemons have an IP in the public network (if used).

• Create a test application Deployment in a new namespace that uses Fusion Data Foundation PVCs.

  Ensure that the test application Pod can write to the mounted PVC volumes. Delete the Deployment's Pod, then wait for it to reschedule to a new node. Ensure that the previous data can be read back, and ensure new data can be written.

• Run some test load generators in the cluster to write to Fusion Data Foundation PVCs. After the Fusion Data Foundation cluster has filled to 20%, induce a hard failure of a node or failure zone. Leave the node or zone in failure state for an hour, and observe the Fusion Data Foundation cluster status and Fusion Data Foundation Pod statuses. Cluster health can be expected to show HEALTH_WARN, but it should not show HEALTH_ERR. One or two Fusion Data Foundation Pod failures may be okay, but several failures could indicate a problem.