From: MERC::"uunet!WKUVX1.BITNET!goathunter" 16-JUL-1992 15:56:29.85 To: supser-l@WKUVX1.BITNET CC: Subj: A couple of bugs.... Hello, all. I hope the Supervisor Series is working well for most everyone! The list has been quiet, so I assume.... There are two problems I'd like to warn you about. These situations are rare, but *can* crash your system. o SUPERVISOR: when supervising a terminal that becomes "hung" (or is "hung" when you start supervising it), the SUPERVISOR process may also hang. If this happens, you should STOP/ID the process running SUPERVISOR *before* you STOP/ID the hung process. At least, I *think* this will prevent a system crash. What's happening is that somehow the target terminal's UCB is getting deleted and SUPERVISOR isn't told that it was deleted. It happily tries to keep accessing a UCB that has been deallocated and is no longer valid, thus the crash. I've come up with a method for trying to catch this case, but I still haven't figured out why SUPERVISOR isn't getting that notification like it should (when the target process is killed, you should get a "Data set hangup" message). o PHOTO: do *not* specify an invalid log file name when using PHOTO. Something is going wrong in the initialization that PHOTO doesn't recover from gracefully. If you do specify an invalid file name (for example, a directory that doesn't exist), PHOTO will continue to work OK for that process, but when it logs out, some PHOTO code goes into an infinite loop that will either hang the system (uniprocessor) or crash it with a CPUSPINWAIT bugcheck (SMP). I know, that makes it difficult to use in an academic environment. I'm trying to track these down and fix them, but my time for Supervisor is very short at the moment (real work beckons). I'll do my best to come up with fixes ASAP. Thanks. Hunter ------ Hunter Goatley, VMS Systems Programmer, Western Kentucky University goathunter@WKUVX1.BITNET, 502-745-5251