You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
When deleting the DRPC we may need to adopt the VRG, delete the
secondary VRG, wait until the secondary VRG is deleted, delete the
primary VRG, and wait until the primary VRG is deleted. This takes 60-90
seconds and many reconciles (18 seen in e2e test), and creates huge
amount of noise in the log.
Suppress the noise by introducing the deleteInProgress error. When the
reconcile is successful but it is still in progress, we return a
deleteInProgress error describing the current progression. The top level
error handle logs an INFO message and requeue the request.
With this change we will see multiple logs for the secondary VRG:
INFO Deleting DRPC in progress {"reason", "secondary VRG deletion in progress"}
...
And finally more logs for the primary VRG:
INFO Deleting DRPC in progress {"reason", "primary VRG deletion in progress"}
...
Notes:
- We logged errors during finalizeDRPC twice; once as INFO log, and once
as ERROR with a stacktrace when we return error from the reconcile.
Remove the duplicate INFO log.
- The linter is not happy about the new nested if. We can avoid this by
extracting a helper to handle finalize errors, but I want to keep the
change minimal for easy backport. We can improve this later upstream.
Signed-off-by: Nir Soffer <nsoffer@redhat.com>
0 commit comments