cluster-manager is not started

[masa213f]

Describe the bug

cluster-manager is not started.

Environments

• Version: MOCO v0.16.1

To Reproduce

1. Create a MySQLCluster
2. Wait for the MySQLCluster to become HEALTHY.
3. Edit the MySQLCluster so that a reconciliation error occurs.

   $kubectl edit mysqlcluster <MySQLCluster>
   ...
   spec:
     mysqlConfigMapName: not-found
     podTemplate:
   ...
   

4. Restart the moco-controller

   $ kubectl rollout restart -n moco-system deployment moco-controller 
   

After the above steps, a cluster-manager (goroutine) for the MySQLCluster will not start.

Expected behavior

After restarting moco-controller, the moco-controller should start cluster-managers for the existing MySQLClusters.

[ymmt2005]

@masa213f Is this a regression?

[masa213f]

@ymmt2005 Probably, this defect has existed for a long time.

[ymmt2005]

@masa213f
Ah, I overlooked this.

Edit the MySQLCluster so that a reconciliation error occurs

Yes, it's currently the way it does. Maybe we should add more validation.