elasticsearch cluster_error ai_generated true

ClusterHealthResponse: status=RED, unassigned_shards=5, reason=ALLOCATION_FAILED

ID: elasticsearch/shard-allocation-failed

Also available as: JSON · Markdown
78%Fix Rate
85%Confidence
3Evidence
2023-01-01First Seen

Version Compatibility

VersionStatusIntroducedDeprecatedNotes
8 active — — —

Root Cause

Cluster red or yellow due to unassigned shards. Disk full, node left, or allocation filtering misconfigured.

generic

Workarounds

  1. 92% success Check allocation explain API for root cause 
    GET _cluster/allocation/explain?pretty  # shows exactly why shard is unassigned

    Sources: https://www.elastic.co/guide/en/elasticsearch/reference/current/

  2. 88% success Free disk space below watermark thresholds 
    Default watermarks: low=85%, high=90%, flood=95%. Free disk or adjust cluster.routing.allocation.disk.watermark settings
  3. 85% success Reroute with retry_failed to reattempt allocation after fixing root cause 
    POST _cluster/reroute?retry_failed=true  # after resolving disk/node issues

Dead Ends

Common approaches that don't work:

  1. Force-allocate primary shards with accept_data_loss 88% fail

    Force allocation of stale primary shards risks losing acknowledged writes and creating split-brain data

  2. Restart all nodes simultaneously to fix red cluster 82% fail

    Simultaneous restart causes full cluster outage and risks data loss if quorum is lost