Amanda-Users

Re: Make amanda unload tape drive?

2003-01-22 10:38:02
Subject: Re: Make amanda unload tape drive?
From: Simon Young <simon-amanda-users AT blackstar.co DOT uk>
To: amanda-users AT amanda DOT org
Date: Wed, 22 Jan 2003 15:12:07 +0000
On Wed, Jan 22, 2003 at 09:28:12AM -0500, Jon LaBadie wrote:
> 
> The original OP did not specify a changer and I, perhaps incorrectly
> formed my answer based on a single drive with no changer.

That's what I thought, but I was hoping it would be just as easy :-)

> My experience with changers is pretty limited and the changer script I
> use (chg-mtx) does not maintain the state of the current slot in a
> disk file.

I just use a straightforward chg-multi with three external drives. The
state of the changer is stored in a state file.

> But based on that limited experience I find that amdump leaves the
> last tape used in the drive. It leaves it at the end of media, after
> the last file is written; i.e. not rewound. The changer script is not
> called again at this point. From this I don't see where the current
> slot state would be incremented past the last tape used.
> 
> Your experience may differ, if so I'd like to know for future
> reference.

Well, I'm quite new to amanda, but I've been running amdump for the last
few nights. Last night's backup was to the tape in slot 2. If I run
'amtape <config> current' now, I get the tape in slot 3. Therefore I'm
assuming the state file must have been updated after amdump ran.

> Again my lack of experience causes this question. Do other changers or
> libraries work differently? Can some eject multiple individual tapes?
> Curious minds would like to know.

With chg-multi, all tapes in the 'changer' can be controlled
individually, since they're really just individual drives.

I was toying with the idea of finding out which is the current slot
*before* runing the backup, then ejecting the tape in this slot
once amdump is finished - but this doesn't work if amanda used more
than one tape.

Any ideas would be appreciated.

Thanks,

Simon.