Problem Definition

Cache replication logging (LocalQCacheManager/LocalQCacheReplicator) is currently output to the atlassian-jira.log. This is undesirable as cache statistics messages are generated every minute, for every node, which overwhelms the logs, particularly when you're looking for non-cache-related events.

A similar situation maintains for index stats logging (snapshot stats, lucene-stats).

Suggested Solution

Separating cache replication and index stats logging to dedicated log (e.g., atlassian-jira-stats.log) would greatly reduce the noise in atlassian-jira.log. Additionally, this will allow for retaining more cache stats and indexing stats logging, which is useful for examining trends over a large time frame.