Rolling Restart of a Coherence Cluster Causes a Reset of The Expiration Time of the Cache Entries
Last updated on JUNE 01, 2017
Applies to:Oracle Coherence - Version 3.7.1 and later
Information in this document applies to any platform.
Coherence cluster is made up of 72 cache servers running on 6 Linux Red Had 6.4 machines, and containing several distributed caches with a configured expiry-delay for their entries (there's several millions of entries), it seems that a rolling restart operation on the whole cluster causes a "reset" of the expiration time of the cache entries, meaning that their expiry-delay restarts elapsing from the time in which the rolling restart is performed, as if all the entries were put in cache at the same time. This causes a MASSIVE eviction of entries (i.e. millions of entries) to happen exactly N days after the rolling restart, where N is the duration (in days) of the configured expiry delay.
All cache entries expire at same time after rolling restart. How to preserve the expiry from before the rolling restart?
Sign In with your My Oracle Support account
Don't have a My Oracle Support account? Click to get started
My Oracle Support provides customers with access to over a
Million Knowledge Articles and hundreds of Community platforms