With really nasty transient problems under MVS, I would set up PER traps in the hardware to catch it the next time around.
VM does it much better with CP TRACE, but it's been a long while since I've done that. Anyway when I get desperate enough to dig out why I'm getting a message from Linux, I would: Find where writes to stderr are handled -- backtrace to the caller -- dump the code, regs and storage -- find the program name and offset -- maybe there's some comments in the source... It would even be possible to collect the data for a message to module table. George Haeh 416-465-2292 [EMAIL PROTECTED]
