Open
Description
Currently, as the cluster topology changes we lose track of the events leading to the change in behavior. I would like to understand if users/operators could benefit from storing events regarding change in cluster state and serve the information similar to COMMANDLOG GET.
- Node marked as failed
- Node marked as healthy
- Node failover
- Node step down
- Slot added/removed/transferred
Alternatives considered
Parsing through logs after incident or capturing cluster nodes output on regular interval from a separate tool and analyzing it.