Annoying "Setting up for writing, done(full)" errors

[coojee]

Ok, I'm using Networker 5.1(Build 298 Extended/130), on a server running HP-UX 10.20 as an operating system. The server has a StorageTek 9714 Jukebox attached which has 4 DLT7000 drives in it.

We regularly get those annoying "Setting up for writing,done (full)" error messages, which make one of the DLT drives totally unusable. The only way to get the drive back in action that I've found is to completely restart Networker. Invariably this error happens right in the middle of the backup schedule, when you least want to restart it!

Does anyone know a method of getting the drive back in action without restarting Networker? People in my company seem to think an ex-employee once found a way, but of course he never documented the fix! Any suggestions would be GREATLY appreciated.

 

[Guest_imported]

I remember something like this. At that time, I did not get the "writing, done" but a "writing, idle". If you give the nsrmmd corresponding with the drive a kill -HUP, the nsrmmd will respawn itself and starts working again.

Ben

 

[coojee]

Thanks for the info. I tried this, but unfortunately it just killed the nsrmmd for the affected drive, and didn't restart the nsrmmd or clear the error. One thing I also forgot to mention is that when we get this error, the jukebox usually (though not always) grinds to a halt. What seems to happen is that Networker becomes incapable of unmounting tapes from other drives, so no new tape load requests are processed.

In case this is useful to anyone, the way to get the Jukebox working again (but not the &quot;Setting up for writing, done (full)&quot; drive) is to either do a hardware reset, or if there are other drives still in use, issue a manually unmount of the drive in the background (nsrjb -f /dev/rmt/<drive> -u &), and if you get a &quot;The Jukebox is currently in use by another process&quot; message, kill an existing load request (e.g. /opt/networker/bin/nsrjb –s Hercules.ape.apci.com –j rd=hades). You still need to restart Netwoker to get back the &quot;Setting up for writing, done(full)&quot; drive though.

If anyone else has found a way of getting a &quot;Setting up for writing, done(full)&quot; drive without restarting Networker it would be very helpful. Thanks.

 

[coojee]

Sorry, the above wasn't entirely correct. The nsrmmd for the drive did eventually restart, but it took 35 minutes to do so. Drive is still out of action though.

 

[DinoCaira]

I have a similar set up to yourself and have experienced the &quot;Setting Up For Writing, Done, Full&quot; Error. I haven't yet found the reason for it occuring but have found the following to cure it:

Find out the DLT's being written to when this error occured.
Then (if you can afford to lose this backup)...
re-label the tape(s) in question.

Re-start the backup group and you should find it will pass.

 

[1407]

What I would really recommend, is to Upgrade your NetWorker to 5.5-003, If you are unable to go directly to 6.0.1.
This use to happend with releases pre 5.5.

I know STK machines and NetWorker. It is a very good combination. It should work nicely.

 

[Guest_imported]

This error message will appear nearly always after the following error:
&quot;Unknown error 1104&quot;

This basically means that the tape you are trying to use was interrupted during its previous backup (e.g the server was rebooted or the services stopped in the middle of a Save session). You need to kill all nsrjb.exe processes and re-label the tape in question.