☂️ Track MCM IT related issues

[rishabh-11]

How to categorize this issue?

/area testing /kind epic /priority 2

What happened:

• [x] https://github.com/gardener/machine-controller-manager/issues/853
• [x] #675
• [x] #789
• [ ] Other improvements
  • Debugging IT is tough right now, the tests should specify which machine objects they created
  • Check the status of machine also once node object is confirmed, because it can happen node obj is there, but machine is still Pending but tests are still passing.
  • Only run the vendored version of MCM during IT in pipeline
• [ ] Enhance IT with tests for the following logics - Bootstrap token creation and deletion should be checked - Unhealthy machine is being marked Unknown should be checked - Draining mechanism should be checked - Meltdown mechanism should be checked - Safety controllers should be checked
• [ ] Validate MCM with IT
  • Currently MCM doesn't have any integration tests. Also we don't gaurantee in pipeline tests that it would not have any runtime errors.
    • Possible soln: A small kind cluster with machine crds installed , then we could start MCM for that and confirm Processing machineDeployment … log is coming.

[himanshu-kun]

Post Grooming Discussion

• We see that we just wait for deletionTimestamp to be put on the machine resource(waiting for delete event), and don't wait for resource to be actually deleted from etcd
• Restarting MC doesn't work for cleanup