OpenSearch High JVM Memory Pressure

How to fix it

// try these in order, the most common fix is first

1. FIX 1Reduce shard overhead

   Copy steps
   1. 01Delete or merge old indices
   2. 02Use shrink or reindex into fewer shards
   3. 03Use ISM policies to remove unused indexes
   Did this fix it? Yes No

2. FIX 2Clear fielddata

   1. 01POST my-index/_cache/clear?fielddata=true (can disrupt running queries)
   Did this fix it? Yes No

3. FIX 3Fix mappings and queries

   Copy steps
   1. 01Aggregate on keyword fields, not text
   2. 02Avoid leading wildcards and huge date ranges
   Did this fix it? Yes No

4. FIX 4Scale

   Copy steps
   1. 01Add data nodes or use a larger instance type
   2. 02Keep max heap at 32 GB or less per node
   Did this fix it? Yes No

Symptoms

JVMMemoryPressure stays high, GC pauses grow, requests slow or are rejected, and on Amazon's service writes are blocked at 92% for 30 minutes.

Why it happens

• C-01Too many shards
• C-02Aggregations or sorting on text fields (fielddata)
• C-03Wildcard or wide time-range queries
• C-04Unbalanced shard allocation
• C-05Instance type too small for load

// this guide also covers

• opensearch jvm memory pressure
• opensearch high heap usage
• opensearch jvm pressure 92%

The same problem in other apps

All slow, laggy or resource-hungry guides →

• Elasticsearch High JVM Memory Pressure and Out of Memory (Elasticsearch)
• Typesense Out of Memory (OOM) and RAM Sizing (Typesense)
• Meilisearch Indexing Slow (Meilisearch)
• Algolia Search Slow or High Latency (Algolia)

Other OpenSearch errors

• OpenSearch Cluster Status Red
• OpenSearch Cluster Status Yellow
• OpenSearch Unassigned Shards
• OpenSearch OutOfMemoryError and Circuit Breaker
• OpenSearch Exceeded Maximum Shard Limit
• OpenSearch Index Read-Only: Disk Watermark and ClusterBlockException
• OpenSearch 429 Too Many Requests
• OpenSearch Docker Container Won't Start