Handle stale managedPDB definitions

[michohl]

In the reconcile loop there are cases where we are attempting to update a managed PDB that has already been updated in the cluster by something else. This results in the following error:

time="2024-07-17T14:20:46Z" level=error msg="Failed to update PDB: Operation cannot be fulfilled on poddisruptionbudgets.policy \"my-cool-deployment-pdb-controller\": the object has been modified; please apply your changes to the latest version and try again"

This error is mostly harmless since another run of the reconcile loop will usually resolve the error but this change will reduce the noise from the error messages and speed up the time to reconcile PDBs which might have needed multiple reconcile loops to finish their update before.

I have been running this change in our own clusters for a week now and since then we have not received any more of the above errors.

[mikkeloscar]

👍

[szuecs]

👍

[szuecs]

thanks @michohl !

[szuecs]

@mikkeloscar I think we need to tag it. Do you do this manually?