I just did my customary sipsorcery test call after getting up this morning and it failed, damn. Terminal serviced into the sipsorcery server and the App Server was running with around 530MB of memory (way too much indicating a leak), 55 threads and 22% CPU. The memory and the threads are nothing new and they used to go a lot higher. However the CPU has never flat lined like that before.
I saved the logs and will trawl through them after work. My suspicion is that having removed the mechanisms to interrupt the execution of dialplan scripts to try and get rid of the memory leak and inconsistent dialplan processing a dialplan has been used that was long running and did need to be cancelled. Probably it was someone inadvertently putting their script into an infinite loop. To be honest I expected it to happen but was hoping for it to be later rather than sooner.
It’s back to the drawing board over this weekend to try and come up with a different solution for dial plan execution interrupt.
Sigh, sigh, sigh….