Monitor health and capacity

Monitor both customer workloads and managed-service readiness. A green Azure resource state does not prove that ingestion, search, or every Elasticsearch index is healthy.

A monitoring agent in the managed resource group forwards operational platform telemetry to the iVedha monitoring cluster. iVedha uses this telemetry to monitor supported platform health. Customers remain responsible for verifying workload outcomes, data quality, ingestion completeness, dashboards, and application-level service objectives.

Review cluster health

With an appropriately privileged user:

curl --fail-with-body --user "<user>" \
  "https://<elasticsearch-hostname>/_cluster/health?pretty"

Review:

• cluster status and unassigned shards;
• node availability;
• disk use and growth;
• indexing rejection or latency;
• search latency and errors;
• shard count and size;
• snapshot failures;
• Kibana and client availability.

Respond to disk pressure

Elasticsearch uses disk watermarks to protect nodes. At the flood-stage watermark it can block writes to affected indices. Do not permanently raise watermarks to hide insufficient capacity.

1. Confirm which nodes and indices are affected.
2. Stop avoidable ingestion bursts.
3. Delete data only under an approved retention decision.
4. Request a supported capacity or topology change when growth is durable.
5. Confirm that write blocks clear and ingestion recovers.

See Elastic watermark errors for current product behavior.

Request a capacity change

Provide current capacity, observed growth, retention, peak ingestion, query load, shard metrics, business deadline, and the requested outcome. Do not ask for a size label without supporting workload evidence.