Will Cluster Services Be Affected If the Usage of a Single Node Is Too High?

Symptom

According to the cluster monitoring information, the disk usage of an Elasticsearch cluster exceeds 80%. Does it affect cluster performance?

Impact on Services

• If the disk usage for a node exceeds 85%, the cluster will not allocate new shards to the node.
• If the disk usage for a node exceeds 90%, the cluster will migrate the node shards to other data nodes with low disk usage.
• If the disk usage for a node exceeds 95%, the read_only_allow_delete attribute will be enabled in its indexes. In this case, indexes on the node can only be read or deleted but do not support data write.

If the usage of a single node is too high, you can scale out the cluster by adding more nodes to the cluster or expanding the capacity of existing nodes. Indexes are not allocated to new nodes immediately. You can open the Cerebro file to check the index allocation of the nodes. You can also change the values of indices.recovery.max_bytes_per_sec and cluster.routing.allocation.cluster_concurrent_rebalance to increase the index allocation speed.