Node out of memory
A node is under severe JVM memory pressure: heap use stays very high while garbage collection runs for much of the time. The node might slow down, reject work, or disconnect.
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.
We detected a high GC overhead on node es-data-01 and es-data-02. The current heap utilization is 30%. GC ran for more than 50% of the time over the last 30 seconds. The node doesn't have much heap memory left and might become unresponsive.
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.
Nodes in Elasticsearch can run out of memory due to various reasons such as insufficient heap memory allocation, large data size, heavy querying or aggregations, mapping and field data issues, concurrent operations, shard overallocation, memory leaks or bugs, and insufficient system resources. To address these issues, you can increase heap memory allocation, optimize mappings and queries, scale horizontally, monitor cluster performance, identify resource-intensive operations, review logs for memory leaks, and consider upgrading hardware if needed.