This is a clone of issue OCPBUGS-56021. The following is the description of the original issue:
—
Description of problem:

When Sriov Operator plans for a node drain, it updates the node annotation with "sriovnetwork.openshift.io/state=Drain_Required". If the user modifies this label to "Idle" before the drain starts, sriov operator does not try to drain the node and stucks in InProgress state. Sriov Operator needs to relabel the node to Drain_Required and proceed with Draining. 

Version-Release number of selected component (if applicable):

How reproducible: 100%

Steps to Reproduce:

• Create SriovNetworkNodePolicy, SriovNetwork and test pods with secondary interface set to Sriov VFs
• Create SriovNetworkPoolConfig with MaxUnavailable set to 1
• Delete the SriovNetworkNodePolicy
• Verify SriovNetworkNodeState has InProgress state for all nodes and only 1 node drains
• Meanwhile update the node annotation of the worker which is yet to be drained with sriovnetwork.openshift.io/state=Idle. This node continues to be in InProgress state

Actual results:

Sriov operator does not drain the node and stuck in InProgress state

Expected results: 

Sriov Operator needs to relabel the node to Drain_Required and proceed with Draining. 

Additional info: