Please help !!! URGENTLY

could someone tells me what should I do, here is the error message I get:

May  1 09:22:45 2000: Backup Server: 4.4.2.1: Device dev/rmt/0n does not belong to session <52>.  Notification failed.

Because of this error, I redo the backup, and it comes the worse, here is another message:

1> dump database master to tapedump with init
2> go
WARNING: In order to LOAD the master database, the SQL Server must run in single-user mode.  If the master database dump uses multiple volumes, you must execute sp_volchanged on another SQL Server at LOAD time in order to signal volume changes.
Backup Server session id is:  64.  Use this value when executing the 'sp_volchanged' system stored procedure after fulfilling any volume change request from the Backup Server.
Backup Server: 4.55.2.1: Device validation error: couldn't open raw device /dev/rmt/0n, error: Device busy
Msg 8009, Level 16, State 1:
Server 'WRDBQ1', Line 1:
Error encountered by Backup Server.  Please refer to Backup Server messages for details.

thanks

Try running (as root):

/etc/fuser -u /dev/rmt/0n

This should identify the process that is holding the drive open.  It's very possible that it's another Backup Server process that's holding the tape drive open.  ie, when you tun ps and grep for the PID number returned by fuser, the program name begins with "nsr".

If this is the case then you can shutdown the Backup Server and restart it.  Login as root, then:

sh /etc/init.d/networker stop
ps -efa ¦ grep nsr

Manually kill any "nsr" processes that are still hanging around.  Then:

sh /etc/init.d/networker start

It may be that the "nsr" processes will not die.  If this is the case, I'm afraid you'll have to reboot the system :(

Hope this helps.

Oops - forgot to say - if fuser returns a non-"nsr" process (for example, a tar command), just try killing the command.  It's still possible, though, that the process won't die and you'll need to reboot if this is the case.

Thank Andy

I think your last message matches with my DBA's decision.  She finally has to reboot all the servers related to that backup server and now everything seems fine.

Thanks anyway, hope that I have a chance to try your solution next time.

(in fact, I kill the ps of backup and when I check the device - mt status - I still get message of 'Device busy' and that is why I consider unload all tapes, power cycle the tape drive, and I still get 'Device busy' ! but according to the message, I already think about reboot the backup server, but I don't have power to do that yet !!! and I never do that before to experience it ! So I end up here :()

CT

Ah, well, there's a few processes that will be running that could still be hanging on to the tape device.

The Backup Server software is Legato Backup Manager licensed to Sun.  We've been using Backup Manager for a couple of years now, and we back up 50 servers (Unix, Novell, and NT) to a single Unix server connected to two DLT drives.  The normal processes we see are:

/bmanager/nsrd          
/bmanager/nsrexecd      
/bmanager/nsrexecd      
/backup/sbin/nsrindexd  
/backup/sbin/nsrmmd -n 2
/backup/sbin/nsrmmdbd   
/backup/sbin/nsrmmd -n 1

(Path names are a little different to the Solaris version.)

So, you might kill off, for example, nsrexecd, but nsrmmd is still there hanging on to the tape drive.

You can restart the Backup Server in two ways.  The way I described above ("sh /etc/init.d/networker stop", kill any left over processes, "sh /etc/init.d/networker start"), or you can just run "sh /etc/init.d/networker restart".

We've found that the "stop, kill, start" sequence works best though, but your mileage may vary :)

Glad to be of help, and hope this helps in the future. :)

Thanks alot Andy

I think I will try next time, the only ps I kill at the time is the backup process, I don't know the others.
Thanks for your advice. Definitely I will try next time.
CT