Forum Discussion

Lylex's avatar
Lylex
Level 3
13 years ago

media eject problem

I'm a substitute SysAdmin here, with a couple of questions on ejecting tapes.  We're running Netbackup 7.0 on Solaris, with a StorageTek L180.

I select a couple of tapes and choose eject.  The eject dialog screen pops up with the selected tapes.  I click "eject", and they do eject.  But after removing the tapes from the MAP, I click "close" but the window just hangs.  I can't close the window, select the Errors tab, or access the main interface anymore.  The only way out is to stop the server processes (/etc/init.d/netbackup stop).

In another case, when I eject a tape to the MAP but don't remove it right away, after a while it gets replaced in the library.  I suppose this is the "Media ejection timeout period"  described in the Unix Server doc.  It probably happened after 30min, though it's a little surprising since the manual only lists a 30min timeout for 8MM and DLT, not LTO.  I'm new to NetBackup, but other backup systems just eject a tape...period.  So I guess I don't understand the design philosophy yet.

Finally, I know the normal SysAdmin pulls the ejected tapes from the MAP, then loads scratch tapes in the MAP before closing out the eject window in the user interface.  The scratch tapes wind up being checked in.   This certainly seems to work, but I'm used to a seperate inventory process to check-in scratch tapes.  What's the normal way to do this under NetBackup?

Thanks...Lyle

  • two step - just to make sure you are doing it right.

    Proper- eject - tells you to remove the tapes from the map and asks you do you want to proceed - and you say yes -  this only works if you ACTUALLY remove the tapes from the MAP.  You then have to give it time to scan the MAP to make sure all tapes have been removed... This might be your close issue - give it time to rescan the MAP.

    ----------------------------------------

    How must people do it.  

    Select your tapes

    say eject

    it moves the tapes to the MAP

    it asks if you want to continue (if you say yes but do NOT remove the tapes from the MAP it will put the back)

    say NO.  give it a little bit to come back and say eject successfull for each tape.

    Now any time you want go remove the tapes from the MAP and at the same time you can refill the MAP with scratch tapes.

    Now go back to console and do the inject.

6 Replies

  • 1st Point

    Try  the command line version - lets se if this works ...

    vmchange -res -multi_eject -verbose -rn 0 -rt tld -ml <media1:media2> -single_cycle -w

    (rn = robot number / -rt = robot type  )

    If it still hangs, will need to look in the volmgr/debug/robot log

     

    2nd point - I believe this is normal, I will be honest, I very rarely eject tapes and when I do I use the command line.  I don't actually have a library I can test this on, else I would - only got a VTL.

    Real pain when you try to eject a VTL tape, you end up with a pile of 1s and 0s on the floor ....  :o)

     

    3rd point

    If you use the command line, this won't be an issue - but again, I believe this is normal.

     

    If in /usr/openv/vomgr/vm.conf you put the line :

    AUTO_UPDATE_ROBOT

    ... and restart the services then (it is depenndant on the library) but the act of simply opening the CAP, putting in a tape and closing it will inject the tape AND run an inventory.

    (only if the library supoorts this)

    Martin

     

     

     

  • Sounds like a bug I came across with media eject. It was a while ago so details are sketchy...I recall it was only the GUI, so manual eject worked , or would prove it one way or the other. They send me some java to fix it. You could try an upgrade to 7.1 which is where Im at.

    Jim

  • Yes, that was my thoughts - sounds bug like - hence why I wanted to see if the command line worked ...

    If it is a bug, you have x3 choices ...

    1.  Log a call with Symantec and see if there is an EEB available

    (I'm busy at the tic, but if I get some time later, I will see if I can find one)

    2.  If the command line works, use that instead

    3,  Try an upgrade

     

    Martin

  • two step - just to make sure you are doing it right.

    Proper- eject - tells you to remove the tapes from the map and asks you do you want to proceed - and you say yes -  this only works if you ACTUALLY remove the tapes from the MAP.  You then have to give it time to scan the MAP to make sure all tapes have been removed... This might be your close issue - give it time to rescan the MAP.

    ----------------------------------------

    How must people do it.  

    Select your tapes

    say eject

    it moves the tapes to the MAP

    it asks if you want to continue (if you say yes but do NOT remove the tapes from the MAP it will put the back)

    say NO.  give it a little bit to come back and say eject successfull for each tape.

    Now any time you want go remove the tapes from the MAP and at the same time you can refill the MAP with scratch tapes.

    Now go back to console and do the inject.

  • 1) What is your Netbackup version?

    I know in 6.5 it may still have some bugs, but in 7.x this vault is pretty stable and we don't get a lot of nbvault process hanging or crashing issue.

    2) Media ejection timeout is to set a timeout, so if media in MAP can't be cleared off in time, Netbackup would know when to stop the "waiting" and reset things back. It is just a status check I believe. 

  • Well it was pretty much "operator error".  Between the browser watching the library and the eject dialog page, the window asking "do you want to continue" got covered up.  Everything does work as expected.

    Thanks for all the speedy and helpful replies....Lyle