Greetings all;
I can recall when amanda would stay within 98% of filling a 4GB tape, and do
it reliably for weeks at a time.
But recently, the size of the backups has been varying more, like by as much
as 30% from nightly run to nightly run. I added a day to the schedule about
2 weeks ago because it seemed to be continually leaving 3 or 400MB in dumps
for the next nightly run to flush.
That should have brought the average fill ratio down to 70-75%. But last
night it did level 0's on 3 of the 4 biggest dle's, filled the vtape and quit
with 709MB left in /dumps, which was from 28 of the 31 dle's here. So I just
added yet another day to the dumpcycle and runspercycle.
However amadmin balance's output is till a bit scary:
[amanda@coyote ~]$ amadmin Daily balance
due-date #fs orig MB out MB balance
----------------------------------------------
9/01 Sat 14 7724 3948 -46.1%
9/02 Sun 1 520 259 -96.5%
9/03 Mon 4 13093 10921 +49.2%
9/04 Tue 6 12916 9176 +25.4%
9/05 Wed 10 24545 12292 +67.9%
----------------------------------------------
TOTAL 35 58798 36596 7319
DISTINCT 31 56718 35560
(estimated 5 runs per dumpcycle)
(1 filesystem overdue, the most being overdue 1 day)
What I don't grok is that the 9/01 data, which was last night, totally
disagrees with what it actually did. Which was 11.87GB & 98.5%, leaving
709MB in /dumps. Should this first line of a balance report not be a
truthfull answer since its after the actual fact?
This is amanda-2.5.2p1-20070727.
Is there an option I can add to the build config, or to my wrapper scripts,
that would improve its estimate accuracy and make it be more consistent?
Or is it time I make the jump to the 2.5.3-alpha code since it appears this is
the end of the development line for 2.5.2, there having been no activity
there for over a month?
--
Cheers, Gene
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
I'll never get off this planet.
-- Luke Skywalker
|