> Yesterday's backups gave me the following error on alot of
>filesystems:
>"stanford /etc lev 1 FAILED [dumps way too big, must skip incremental
>dumps]" without any reason what so ever. The total size of this
>specific filesystem is: 1279 KB!!! Why did amanda decide that this
>filesystem is too big and it must skip the incremental backup???
Amanda first picks what it would like to do for this run and compares
the estimated size against the tape size (length from the tapetype times
runtapes). If the estimated size is too large, it will start deferring
full dumps to incrementals until it will fit. If it does everything it
can and the total estimated size is larger than a tape, you get the above
message and it starts skipping backups based on (as I recall) priority.
You need to look at your amdump.<nn> file and find the tape size Amanda
thought it had to work with and what it started with for estimates.
Look for the "DELAYING DUMPS IF NEEDED" line. The "total_size" value
is the estimated size to that point. The "tape length" value is the
tape length. You need to subtract from that the number of file systems
to back up times the "mark" value.
Look also at the "ANALYZING ESTIMATES" section and see what Amanda
thought it had been told to do. For instance, forcing it to do a lot
of full dumps might cause trouble.
> Paolo
John R. Jackson, Technical Software Specialist, [EMAIL PROTECTED]
