In a cache loader function(com.atlassian.cache.ehcache.LoadingCache), one hanging database request has a chance to block all other cache loader functions, that use the same cache key partition. This can cause blocking cache to affect other parts of the application.

It would be reasonable to add a default timeout or a configurable timeout while loading the caches so that this doesn't cascade and cause degradation of performance in Confluence. Instead of calling a lock(), it makes sense to user trylock() with a timeout