we have been monitoring our weblogic 10.3 cluster members for a couple of months using a wlst script which pulls the ThreadPoolRuntime.QueueDepth measurement from each server in the domain. Since patching to 10.3.6 we've noticed that many of our servers have been returning a steadily growing number for this metric ( > 1100 ) after a week of up time. Doing some reading the QueueDepth number includes system requests and pending user requests and the separate PendingUserRequests number is mostly at 0. Any ideas what the system requests are that are not being serviced or still pending? Or any idea how to find out what these are?