Pivotal Knowledge Base


Monitoring - Details on how to interpret statistic metrics for TTL and LRU

Applies to



The purpose of this article is to explain how to interpret the metrics that exists for TTL expiration and LRU eviction.



For LRU there is a separate metric called lruDestroys in the statistics. This metric does not show TTL destroys. But in addition to LRU destroys it also shows normal region.destroys() from Java API or GFSH CLI calls, which can be somewhat confusing. So in effect you can have lruDestroys showing greater than zero while no entries have been destroyed by LRU eviction. In this situation the lruEviction will show zero while lruDestroy will be greater than zero.


TTL expiration does not have a special metric category in the statistics like LRU does. Instead TTL is shown in the generic destroy metric on region level. In the statistics these are in: CachePerfStats - RegionStats-partion-[region name]. This metric shows both region.destroys() and TTL destroys.


Powered by Zendesk