We see in the automated regression tests that eventually the worker MCP is degraded. At the beginning we thought that it was due to the drain operation not being successful, but even if it  can be related to the degradation we could not reproduce it by only making the drain operation fail.

  In the no-techpreview regression tests we don't see this behaviour.

    

4.18
    

Intermittent 
    

    1. Unfortunately, we don't have the steps to reproduce this issue. It happens in our regression tests but we don't know yet how to exactly reproduce it.
    2.
    3.
    

The worker MCP is degraded and no reason is reported. Apart from the drain operation reporting a failure in the MachineConfigNode status we don't see anything else that could cause the degradation. Nevertheless, making the drain fail is not enough to reproduce it.

    

No degradation should happen, and if a degradation happens a clear reason for this degradation should be reported.
    

We include the must gather file and the machineconfignode resource for the degraded node.