On Sat, Dec 11, 2010 at 11:12:44PM +0100, Marc Muehlfeld wrote:
> Jon LaBadie schrieb:
> > Not certain I understand the question(s).
> 
> What I mean:
> 
> When amanda is working here, first a lot of the DLEs that have been dumped
> are put on the holding disk first, while amanda is dumping other DLEs.
> Later amanda is putting them to tape. Why not earlier?
> 
> In my amstatus output you can see, that 95 of 121 DLEs had already been
> dumped. 1 of the 95 was allready put to tape. The other 94 DLEs are still
> in the holdingdisk at the time I grabbed this output. Later they are all
> send to tape.

Are you certain that the taper was idle during this period?  Might one
DLE being dumped directly to tape?  If that DLE took a long time to
dump (and tape) then all the other DLEs could be doing is collecting
on the holding disk.

I've not used them, but several amanda.conf parameters could cause the
behavior you see.  They specify to not begin taping until a condition
is met.  For example, flush-threshold-dumped set to 75, will not tape
until the amount of completed DLE data available on the holding disk is
enough to fill a tape at least 75%.

> When I directly dump to tape, only one DLE can be backuped at the same time.

DLEs should still be able to be dumped to the holding disk at the same
time as the one going direct to tape.

-- 
Jon H. LaBadie                  [email protected]
 JG Computing
 12027 Creekbend Drive          (703) 787-0884
 Reston, VA  20194              (703) 787-0922 (fax)

Reply via email to