High replica count on cold nodes
Some indices on cold nodes have a higher replica count than the configured threshold. Extra replicas duplicate every primary shard and still consume storage and cluster resources on cold nodes that typically hold rarely queried data.
For a complete list of insights, refer to AutoOps insights.
| Field | Value |
|---|---|
| Component | Elasticsearch |
| Severity | Low |
| Scope | Node |
| Domains | replication, resource-utilization |
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.
2 replicas on cold nodes for these indices: logs-prod-000045, logs-prod-000046.
AutoOps shows different recommendations depending on how their conditions match your deployment or cluster.
Reduce replicas to zero
Condition: Shown when replication factor exceeds cold tier threshold.
Confirm logs-prod-000045 is backed up, then set number_of_replicas to 0 on the cold tier with the action below (currently 2 replicas). Fewer replicas can increase search latency because fewer nodes serve reads.
PUT logs-prod-000045/_settings
{
"index": {
"number_of_replicas": 0
}
}
Requires the manage index privilege. Requires Elasticsearch 8.0.0 or later. This action changes cluster or index configuration.
Each replica is a full copy of a primary shard. Higher replica counts improve availability and search capacity, but they also multiply disk use and write amplification: every index, update, and recovery must reach every replica.
On the cold tier, which usually holds rarely searched data, more replicas than you need for fault tolerance still occupy storage and recovery capacity. If the replica count stays high without a matching availability requirement, cold-tier cost rises while search benefit stays limited.