Yes I know the answer is don't make Amanda do that :-) Alas I don't think I can make Amanda do anything else.

I have a growing number 160TB partitions to which completed research data is written in 20TBish batches, annotated, then set as immutable.

I also have a LTO-5 tape library with 100 useable slots so I could just about do a single full backup but only for one partition.

What I'd like to do is have 20 slots assigned to each partition and manually run a full backup when a partition reached ~20TB, then eject, offsite and replace the full tapes. When another
~20TB builds up, repeat the process running an incremental. After 8 cycles I would have full filesystem and a volume of tapes from which I could use Amanda to recover anything from single files to the whole partition. I could then reuse the slots to cover a new partition.

I've Googled round with "amanda archive" and the like, but only found posts that almost(1) but don't quite cover this

Is there a best practice/howto document out there that covers this sort of thing?


(1) FX http://www.backupcentral.com/phpBB2/two-way-mirrors-of-external-mailing-lists-3/amanda-20/archiving-with-amanda-66198/