Apache ZooKeeper

Quorum at Risk - Synced Followers Below Safety Threshold

critical
reliabilityUpdated Feb 6, 2026

ZooKeeper ensemble health depends on maintaining a quorum of synchronized followers. When synced_followers drops to exactly quorum size, the cluster is one failure away from losing write availability.

How to detect:

Alert when zookeeper.synced_followers equals the minimum quorum requirement (e.g., 3 in a 5-node ensemble) or drops below quorum. This indicates the ensemble cannot tolerate another node failure without losing write capability.

Recommended action:

Immediately investigate why followers are not syncing: check network connectivity between nodes, verify disk I/O performance on lagging nodes, examine GC pauses in JVM logs, and review zookeeper.log for sync errors. Avoid maintenance operations until all followers are synchronized.