Hi,
First time -users question, I have inherited an amanda installation.
My question might be a FAQ.
I notice that if I run 'amadmin daily info' after a daily run,
I see a bunch of entries I feel comfortable with, like this one:
Dumps: lev datestmp tape file origK compK secs
0 20160120 D03015 13 25990880 25990880 2610
1 20160125 D03001 51 17900 17900 1133
Ok, so this DLE had a full dump on 1/20 and an increment on 1/25
and both are on tape. I like that.
Dumps: lev datestmp tape file origK compK secs
0 20160122 D03018 21 62438810 62438810 10453
1 20160125 0 50230 50230 1048
Here's one where the 1/22 full dump is taped, and the 1/25 increment
is still in the holdingdisk waiting for flush-threshold-dumped.
Right? Eh, ok. The tape alone would get me back to 1/22, anyway.
Dumps: lev datestmp tape file origK compK secs
0 20160124 0 10375270 10375270 291
1 20160125 D03001 67 810 810 18
For this one, am I seeing that amanda's fit algorithm decided to
ship the 1/25 increment off to tape, while the 1/24 level 0
*that it needs to be restorable* is still on the holding disk only?
So, if I had only my tapes, this would be useless to me?
Or am I not reading this output correctly?
How about this one then:
Dumps: lev datestmp tape file origK compK secs
0 20160124 0 20550680 20550680 369
1 20160125 0 680 680 11
2 19691231 D03001 69 0 0 0
Check out the datestmp on that last one. I've seen several
like that, they seem to all have size 0, so maybe when there
has been no change to back up, the empty increment has a zero
datestmp?
So for this DLE, I've got a level 0 and level 1 still waiting to
move off the holding disk, and a level 2 that has been taped but
wouldn't do me any good without them (even if it weren't empty)?
If I am reading these things right, I guess my next question is,
is there some simple configuration option we have overlooked? Is
there a way to have the tape scheduling algorithm try harder to
get the lower-level dumps onto tape ahead of the increments that
depend on them?
This is not the latest version by the way, it's 3.2.3, in case
any of this already works differently in the current version.
Thanks!
Chapman Flack
