Problem Definition

After upgrading to JIRA 8.10.0 and later, including the LTS release 8.13.0, the background reindex operations may be slower (0-20%) than it was in previous Jira versions. The extra cost is caused by higher index consistency ("conditional update" is one of the steps to ensure consistency). The performance overhead is connected with creating multiple Lucene index readers in same request.

See also Background reindex is slow after upgrading to Jira 8.10 and later

Suggested Solution

• Create index searcher once and never invalidate it. This is under assumption that index searcher needs to only be accurate about issues that are not in the cache
• Increase the cache size x100 (capacity 100,000). The searcher never gets invalidated, unless the cache is also invalidated.
  • This means we pay the cost of invalidating the searcher much less frequently.

Workaround

None