Jon LaBadie wrote:
On Mon, Aug 11, 2003 at 11:26:17AM -0600, Orion Poplawski wrote:

Yeah, I would love it if amanda waited until the largest dump that would fit on the tape completed before writing anything to the tape. This would result in a longer dump process, but I wouldn't mind because tape time is small compared to dump time for my setup.


I assume you do not regularly have dumps that do not fit.
Otherwise something should change in your setup.

If that is the case, why not use taper algo "largest"?


I have dumps fairly regularly that don't fit. See my previous posts about problems with size estimates for details. I suppose I could increase my tapecycle but I'm cheap.


taper algo "largest" is even worse than largestfit. Remember, the algorithm only takes into account dumps that are in the holding area at the time sendaflush() is called, namely when a taper becomes idle (I think), not all of the dumps as a whole (which is what I want).




Reply via email to