Loading

Search rejection

The search thread pool queue is full and Elasticsearch is rejecting new search requests on the listed nodes. Clients see errors instead of slow responses, which breaks dashboards and applications immediately.

Note

For a complete list of insights, refer to AutoOps insights.

Field Value
Component Elasticsearch
Severity Medium
Scope Node
Domains performance, search, stability

The following is an example of what you might see when this insight is triggered. Real insights use live data and links from your deployment or cluster.

The search thread pool queue on es-data-01 has reached its limit. Elasticsearch is rejecting new search requests on es-data-01.

Note

AutoOps shows different recommendations depending on how their conditions match your deployment or cluster.

When the search thread pool queue fills up, Elasticsearch rejects new search requests immediately; callers receive error responses rather than waiting. This is the hard-failure stage that occurs after the search queue starts backing up. User-visible effects include failed dashboard refreshes, API errors in applications querying this node, and alert noise from dependent monitoring systems. Both coordinating nodes and data nodes can be sources of queue saturation, depending on where search traffic is routed. If the load is sustained, check whether search traffic can be spread across more nodes or whether queries can be optimized to reduce thread pool pressure. Adding coordinating or data capacity is the longer-term resolution when search volume genuinely exceeds current cluster headroom.