Node evictions in a RAC one node configuration
Hi,
My customer's application is running in a 'RAC one node' configuration - Oracle version 12.1.0.2 under operating system Linux (X86-64). The (3rd party) application is not written in accordance with Oracle' best practices, and when it initiates a large number of separate sessions all accessing the same table cpu consumption rapidly climbs, sometimes hitting 100% at which time a node eviction frequently occurs due to there being insufficient cpu available to respond to clusterware heartbeat prompts. This has been discussed and one solution would appear to be 'instance caging', restricting the database to a subset of the total number