On Sun, Jul 02, 2006 at 01:53:56PM -0400, Scott R. Burns wrote:
> It would appear I have my tape size set to be slightly larger than the tape
> really is (HP DAT24x6 changer). The issue I have now is a dump failed for a
> large DLE and is sitting on disk, this dump is just slightly larger than the
> tape so most likely putting in another tape and flushing it is just going to
> fail again.
> 
>  
> 
> If I lower the tape size in the config, cleanup some files on that DLE
> things will work again but that DLE dump will be stuck on the holding disk
> forever.
> 
>  
> 
> What is the "proper" way to deal with this? Correct my Amanda config
> problems obviously, cleanup the disk so it will fit on one tape, cleanup
> that DLE files from the holding disk, mark that DLE as requiring a full dump
> next run and just let Amanda run normally?
> 
>  
> 
> I am waiting for a bit before I jump onto the tape splitting version as this
> is my production server for the entire company and I am a little shy of
> something so new.
> 

No "standard" way to deal with this situation that I know of.
Several possibilities spring to my mind.

if HD space is ample, leave it on the holding disk til your
normal tapecycle rotation comes around again.  amanda knows
it is there and amrestore/amrecover work from HD file(s).

just to confirm, there weren't other DLE's taped as well
so that it just might fit when flushed

forget about it and after cleanup, force a new level 0 or
split the DLE into two pieces that will immediately get
level 0's or can be forced as need be.

if you are running an amanda version that supports tape spanning,
temporarily abandon your shyness for just the amflush.

Get that DLT or DDS-4 drive from the other room and label one
tape for amanda and do the amflush.  leave your tapecycle alone
and mark the new tape "no-reuse" after the amflush.

others?

-- 
Jon H. LaBadie                  [EMAIL PROTECTED]
 JG Computing
 4455 Province Line Road        (609) 252-0159
 Princeton, NJ  08540-4322      (609) 683-7220 (fax)

Reply via email to