Intermittently, when OCL is disabled and the nodes are updated with the new non-ocl image the process is not honoring the maxUnavailable value in the MCP.

    

4.18
    

Intermittent
    

  
 In the second must-gather file the issue happened when we did:

   1. Enable OCL in the worker pool
   2. Remove the MOSC resource to disable OCL
   
The result was that the worker pool had 2 nodes being drained at the same time.
   
   
   
    
    

The worker pool has 2 nodes being drained at the same time.
    

Since MCP has no maxUnavailable value defined, the default value is 1. Hence, there should be only 1 node being drained at a time.

Moreover, they should be updated in the right order.