Issue Summary

Reindex on OpenSearch consumes a lot of heap memory

Steps to Reproduce

1. Configure Jira with OpenSearch
2. Perform a reindex

Expected Results

Reindex completes successfully

Actual Results

• Heap consumption keeps growing as the reindex progresses
• Eventually could result in OutOfMemoryError
• Inspecting the heap reveals numerous instances of IssueLinkImpl

Workaround

Use a setting introduced in JRASERVER-67502, specifically by adding this JVM argument:
 -Djira.cache.request.bounded.enabled=true

By default, this will limit each cache to 1000 objects, which could be reduced further with the following argument:
 -Djira.cache.request.bounded.size=200

Effect: each cache in RequestCacheContext won't grow indefinitely. The cache will only grow up to 200 entries, then it'll start evicting old entries.

Note that jira.cache.request.bounded.size is per thread. So the actual max size is multiplied by the number of threads, which is configured via jira.search.api.reindex.consumer.threads (default: 40).