Coherence Elastic Data Does Not Compact Flash Journal Space (Doc ID 2482321.1)

Last updated on AUGUST 29, 2024

Applies to:

Symptoms

Customer has reported the below issue on 12.2.1.3.1 version. 

Coherence Elastic Data Does Not Compact Flash Journal Space. Client is using elastic data for several caches. In this particular case, data are located in 6 storage-enabled nodes (located in 3 physical machines), caches are accessed through custom Web Services located in WLS nodes (non-storage enabled members) and data are synchronized with HotCache (non-storage enabled member of the cluster).  When cache data was initially loaded elastic data use approximately 3GB RAM journal and 3GB of flash journal. The problem is that after some time flash journal space gets exhausted although the max configured size is 16GB () and soft limit is 8GB (). This gets worse during restarts. More specifically, if nodes 1 and 2 (which are located on the same machine) are not restarted, but the rest of the storage enabled nodes (3-6) are restarted multiple times. If this happens, flash journal runs out of space eventually (while the amount of data is stable and is approximately 6GB per node with primary and backup included). Other note is that even if user request an explicit full compaction through JMX, only a tiny amount of flash journal space is reclaimed.

Cause

My Oracle Support provides customers with access to over a million knowledge articles and a vibrant support community of peers and Oracle experts.