Would it be more helpful if the thread stack dumps produced by sending a running Mono process a SIGQUIT came out in order, one thread at a time, instead of all mixed up?
I have a socket server application with lots of threads (one per incoming connection, plus a few more), which eventually stops working after about 6 hours (it accepts incoming connections, but does not appear to handle their input). When I send it a SIGQUIT, I get lots of stuff all muddled up - stack trace lines from different threads interleaved, and (I suspect) not even always in the right order (i.e. a caller sometimes appearing before a callee). Now I have a vague suspicion that there is a problem with Monitor and inter-thread locking - I have had threads Waiting on a Pulse which never seem to get the Pulse. Is the same locking used during thread dumping, and could it possibly be going wrong? -- Nikki Locke, Trumphurst Ltd. PC & Unix consultancy & programming http://www.trumphurst.com/ _______________________________________________ Mono-list maillist - [email protected] http://lists.ximian.com/mailman/listinfo/mono-list
