Node near out of memory
A node is under severe JVM memory pressure: heap use stays very high while garbage collection consumes much of the time. The node might slow down, reject work, or disconnect from the cluster.
For a complete list of insights, refer to AutoOps insights.
| Field | Value |
|---|---|
| Component | Elasticsearch |
| Severity | High |
| Scope | Node |
| Domains | performance, memory, 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.
High GC overhead on es-data-01 and es-data-02. Heap utilization is 30%. GC ran for more than 50% of the time over the last 30 seconds.
AutoOps shows different recommendations depending on how their conditions match your deployment or cluster.
Restart es-data-01 node
Condition: Always shown for this insight.
Restart the es-data-01 node to clear abnormal load and recover normal operation.
When heap stays near full and collectors run most of the time, the JVM spends capacity reclaiming memory instead of serving search and indexing. Pause times grow, thread pools back up, and the node can become unresponsive or leave the cluster.
Typical drivers include heap sized too small for the workload, expensive queries and aggregations, large field data, shard over-allocation, and concurrent heavy operations. Sustained pressure raises the risk of out-of-memory failures and cluster instability.