elasticsearch throughput_error ai_generated true

EsRejectedExecutionException: rejected execution of coordinating operation [coordinating_and_primary_bytes=0, replica_bytes=0, all_bytes=0, coordinating_operation_bytes=xxxx, max_coordinating_and_primary_bytes=xxx]

ID: elasticsearch/rejected-execution-exception

Also available as: JSON · Markdown
82%Fix Rate
87%Confidence
3Evidence
2023-01-01First Seen

Version Compatibility

VersionStatusIntroducedDeprecatedNotes
8 active — — —

Root Cause

Thread pool queue is full, rejecting new operations. Indexing or search rate exceeds cluster capacity.

generic

Workarounds

  1. 92% success Implement client-side backoff and retry on 429/rejected responses 
    Catch 429 TOO_MANY_REQUESTS and retry with exponential backoff; bulk API returns per-item status

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

  2. 88% success Reduce bulk request size and concurrent bulk threads 
    Start with 5-15 MB bulk size and 1-2 bulk threads per node; increase gradually while monitoring rejections
  3. 82% success Scale out by adding data nodes to distribute load 
    More shards across more nodes increases write throughput linearly

Dead Ends

Common approaches that don't work:

  1. Increase thread pool queue size to a very large value 80% fail

    Large queues just delay rejection while consuming more heap memory. The real issue is ingestion rate exceeding throughput.

  2. Set refresh_interval=-1 to speed up indexing 65% fail

    Disabling refresh makes recently indexed docs invisible to search and does not address queue capacity