home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.os.vms
- Path: sparky!uunet!ukma!darwin.sura.net!aplcen.apl.jhu.edu!procon!rubin
- From: rubin@procon.jhuapl.edu (Don Rubin)
- Subject: Wierd BACKUP change in VMS 5.5-2
- Message-ID: <1992Nov17.203345.23090@aplcen.apl.jhu.edu>
- Lines: 28
- Sender: news@aplcen.apl.jhu.edu (USENET News System)
- Organization: Johns Hopkins Univ APL
- Date: Tue, 17 Nov 92 20:33:45 GMT
- Lines: 28
-
-
- We recently upgraded one of our clusters to 5.5-2, everything
- went well and we had no complaints from the users until someone
- ran one of my programs that is a front end to BACKUP. We have
- a database of backups that allows the user find which tape has
- the desired data. The restore program gets info from the user
- and then mounts the tape (all of our programs do mounts and
- dismounts so the users never have to worry about doing that)
- if the label agrees with the one the user specified then a
- subprocess is created (using smg$create_subprocess) and BACKUP
- is spawned. UNTIL 5.5-2 this worked properly, since the drive
- was mounted in the parent process, backup just went ahead and
- did the restore. NOW backup aborts saying the tape drive is
- allocated by another process (yes, the parent). After much
- mail with DSNLINK the word is that the problem has been
- refered to engineering. We worked around the problem by
- mounting and dismounting the tape before going to the sub-
- process. The problem with this is if the user control Cs
- the program the exit handler can not dismount the tape (as it
- did before when it was mounted by the main process) and the
- drive can end up allocated to a non-existant process.
-
- Just thought you'd like to know.
-
- --
- Don Rubin
- Johns Hopkins Univ APL
- rubin@procon.jhuapl.edu
-