COK jobs not ending after ENDNET

Hi,

We have an occassional issue where some JDENET_K jobs remain active for over an hour after the ENDNET command is run. It seems like this happens when we have some unusual issues with the COKs throughout the week (BSFN failures of a sort). I haven't been the one to have my hands on during one of these events, so I don't have a whole lot of detail. However, I am the one being asked to resolve it

Has anyone had issues like this? Any thoughts?

Thanks
Doug

Interesting 3 letter shortening of Call Object Kernels. Anyway, this was on XE but I saw kernels hang once when I accidentally ran CLRIPC before all kernels had ended.

You may want to see if you have any zombies. If so, clean them up and see if the kernels end faster.

I figured out what was causing this problem. The ENDNET command seems to hold a mutex that the ending jobs are waiting to be released. I stumbled upon this by I investigating the mutexes and threads for these jobs. I noticed my interactive job that I submitted the ENDNET command from was holding a mutex these jobs appeared to be waiting for. Once I logged off, the jobs quickly started to end.

I don't know much about mutexes and threads. It seems strange that this doesn't happen all the time. It seems to only happen when some other thread communications problem exists with EnterpriseOne. Does anyone have any thoughts to this?