Pending Proposals Accumulating in Raft Queue

warning

Resource ContentionUpdated Feb 9, 2026

Growing etcd_server_proposals_pending count indicates that write proposals are queuing up faster than they can be committed, suggesting performance bottlenecks or consensus issues.

Sources

How to Monitor etcd Latency and Disk IO for Cluster Healthoneuptime.com

Key metrics for monitoring etcd - Datadogwww.datadoghq.com

Technologies:

etcdSymptoms of this issue are visible in etcd metrics and logs

How to detect:

Monitor etcd_server_proposals_pending gauge. Alert when pending proposals remain elevated (>10) for extended periods. Check etcd_server_proposals_failed_total for failed proposals.

Recommended action:

Investigate disk and network latency as these slow down Raft consensus. Check for failed proposals indicating leader election issues. Review API server request patterns for write-heavy workloads causing backpressure. Consider batching or rate-limiting client writes if necessary.