Epic Goal

• This is regarding refactoring we need to do in rbac reconciler to mitigate the slowness happening, which is based on number of namespaces

Why is this important?

• This is impacting the user experience for the operator and also sometime end up in downtime for new namespaces created for the Tekton users.

Scenarios

1. Right now, rbac reconciler is going into loop and processing all the namespaces one by one and i have observed it takes around 40s-50s to reconcile one namespace, which can end up in high number on big clusters.
2. Also on every upgrade to new version, this reconciler runs on every namespace ending up in downtime for any new namespace created during that phase.
3. Because of mis configuration and deletion of resource, if reconciler fails to process  namespace it just gets stuck at that namespace and all reconciling stops.
4. In recent example, it took almost 1day to recover the reconciler in dev sandbox cluster after fixing where it was stuck

Acceptance Criteria (Mandatory)

• Need better parallel reconciling
• Reconciling based on event to not process all the complete list
• Better reconcillation to recreate resources.

Done Checklist

• Acceptance criteria are met