DPU object is in ERROR phase when there is "pre bfb-install" failure, no auto re-provisioning is attempted on timeout

[wpeng102]

DPU object is in ERROR phase when there is "pre bfb-install" failure, no auto re-provisioning is attempted on timeout

Provisioning is stuck on ERROR phase when there is "pre bfb-install" failure, no auto re-provisioning is attempted on timeout. only way to trigger re-provision is to delete the DPU object.

DMS timeout only take effect on "active" API (bfb-install).

In our case the failures happened on inactive API (pre bfb-install) due to:

• old "dpf-operator-system-client-secret" secret is not removed when new DPU object come up

• NFS was not available to the DMS pod when it came up

Mitigating the Issue:

Deleting the DPU object will triger re-provisioning.