How to check cluster health

This guide shows you how to inspect whether a running cluster is behaving normally using the output of dqlite-utils.

Steps

Type and run cd /path/to/dqlite/dir
Type and run watch -n 1 'dqlite-utils -c ".status;.log --compact" | head -n25
Observe the term output for a few moments—it should be stable. If this value increases frequently, it indicates that the nodes are failing to elect a leader or that the elected leader is frequently becoming unavailable.
Observe the current_index for a few moments—it should be increasing at the rate new data is written by the leader. If current_index and term are both constant, it indicates that the leader is not being asked to write data.
Observe the log output (re-run the command from step 1 without --compact for more information). This should corroborate the observations from steps 2 and 3.
Type and run watch -n 1 'dqlite-utils -c .config and check the set of nodes this server believes are in the cluster.
If the problem persists, read how-to-check-data-integrity

Previous Backup a dqlite cluster Next Check data-integrity