>Looking for "DELAYING DUMPS IF NEEDED" in amdump.NN produced:
>
>DELAYING DUMPS IF NEEDED, total_size 43441709, tape length 12019712 mark 392
>planner: FAILED biancha /dev/hda5 0 [dumps too big, but cannot incremental dum
>p new disk]
>planner: FAILED ikarus /dev/hda1 0 [dumps too big, but cannot incremental dump
> new disk]
Note two things here. The total size from the initial estimate pass
is 43 GBytes, which is **way** over your tape size. That's why Amanda
dropped into delaying mode.
And the "dumps too big, but cannot incremental dump new disk" message
says "I've delayed as much as I can, but I've got this new disk (biancha
/dev/hda5) that has never been backed up before and I can't do a full
dump of it".
So clearly something has changed -- you've added at least one new client
or disk.
Look back further in the amdump.1 file and you'll find a bunch of
lines like this:
got result for host gandalf.cc.purdue.edu disk /: 0 -> 515252K, 1 -> 10393K, -1 ->
-1K
This says the estimated size for a level 0 of / on gandalf.cc was 515252
KBytes, a level 1 is 10393 KBytes and planner didn't bother asking about
a level 2.
Look through those lines and make sure some system isn't way out of whack.
A little further on is the "ANALYZING ESTIMATES" section where Amanda
tells you why it makes whatever initial guess it makes and there might
be some good information in there.
>Another weird thing is that, on the Amanda backup reports, the % of tape used
>is always only around 3% and yet Amanda gives me the "No space left on device"
That 3% is the amount Amanda **successfully** wrote to tape. Based on
all of the above, I'm guessing it was writing a very large image when
it banged into EOT. In the NOTES section it should tell you how much
had really been written when it got the error, e.g.:
taper: tape 041408/lookout kb 45660608 fm 32 writing file: short write
taper: retrying lookout.cc.purdue.edu:/home/lookout/u.4 on new tape ...
This says taper had written ~45 GBytes to 041408/lookout when it got
a short write error. The next line says it was working on a level 4
backup of /home/lookout/u from lookout.cc.purdue.edu at the time (you
may not have this line if you don't have a changer configured).
Take a look at what file system was being processed at the time taper ran
into trouble. I'm guessing it's very large (almost a full tape worth),
or else the tape (or drive) is having trouble and not writing as much
as it used to).
You might also check to make sure you don't have both hardware and
software compression turned on.
>Patrick Daigle
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
