fdb-kubernetes-operator icon indicating copy to clipboard operation
fdb-kubernetes-operator copied to clipboard

Reduce IgnoreMissingProcessesSeconds time to 0

Open johscheuer opened this issue 2 years ago • 0 comments

What would you like to be added/changed?

In the bounce controller we currently have this code:

		// Ignore processes that are missing for more than 30 seconds e.mg. if the process is network partitioned.
		// This is required since the update status will not update the SidecarUnreachable setting if a process is
		// missing in the status.
		if missingTime := processGroup.GetConditionTime(fdbv1beta2.MissingProcesses); missingTime != nil {
			if time.Unix(*missingTime, 0).Add(cluster.GetIgnoreMissingProcessesSeconds()).Before(time.Now()) {
				logger.Info("ignore process group with missing process", "processGroupID", processGroup.ProcessGroupID)
				missingProcesses++
				continue
			}
		}

to ignore processes that are missing for more than 30 seconds (default value). We can/should change this default setting to 0 to reduce the wait time for the cluster wide restart.

johscheuer avatar Feb 24 '23 15:02 johscheuer