We're trying to debug a particularly trick problem with a channel exit, and we're finding that the MQ signal handler can make it quite hard to find the actual problem.
On MQ 5.3 I've found that it is possible to: export MQS_ACTION_ON_EXCEPTION=ABORT to get a core created by a call to abort() inside the MQ signal handler, but I think it would be easier to debug the problem if we could turn off the MQ signal handler altogether. We've tried attaching to the running amqrmppa process using gdb, but (of course) the problem doesn't seem to occur when we run like this. Are there any environment variables or commands we could use to help us debug this problem? For example - is it possible to turn off MQ signal handling altogether (or just for the channel processes) so that the core we get is at the time the SIGSEGV happens? Or other things which might help. I ran the strings command against the /opt/mqm/bin and /opt/mqm/lib directories, and noticed the strings AMQ_DEBUG_CHI and MQS_DEBUG_SIGNAL which might be relevant. but I'm not about to poke about with these without at least some guidance. There also appears to be a chin debugger command (amqsrdbgm) but again I'm not about to play with this without knowing what I'm about. Any help from the channels team or MQ tech support gratefully received. Regards, David C. Partridge Security and MQ Products Manager Primeur Group Tel: +44 (0)1926 511058 Mobile: +44 (0)7713 880197 Instructions for managing your mailing list subscription are provided in the Listserv General Users Guide available at http://www.lsoft.com Archive: http://vm.akh-wien.ac.at/MQSeries.archive
