This content has been marked as final. Show 2 replies
I figured this out. When an *Extend client is marked as a suspect, a message is logged on the proxy node. It also mentions the time period further requests from the suspect client will be ignored for.
You can also interrogate JMX to see the queue depths on a per connection basis - each connection having an OutgoingByteBacklog & OutgoingMessageBacklog setting. You can compare these values to the suspect protocol settings, (http://docs.oracle.com/cd/E24290_01/coh.371/e22837/appendix_cacheconfig.htm#COHDG371), to see which connections are market suspect, or are approaching the suspect or cutoff limits.