I just moved an AIX node out of one primary sequential storage pool into
another primary sequential storage pool. As part of the process I did "Move
data" processes for all volumes for that node from the old storage pool to
the new storage pool. The pools were both collocated at the time, so this
was easy. Everything worked fine and 100 percent of the data for that node
moved as expected. All new backups for that node go to the new pool.

Except for one small subdirectory. It has about 8 small files in it. It is
not NFS mounted. It is not a filesystem or filespace unto itself. It is
about 200kb total space in a filesystem/filespace with about 3gb in it. It
is a 5th level sub-directory of the /home filesystem. There are 50 other
similarly named, permissioned and owned sub-directories with the same kind
of content in the parent directory. I can find nothing unique about this
sub-directory. And yet 3 nights in a row, when backing up this node, this
sub-directory insists on going back to the previous storage pool. 150gb of
data each night stay in the correct storage pool. 200kb do not. The node has
a total of over 8 terabytes of data stored in TSM.

BTW we do not have HSM implemented.

The client is 4.2.2.1, the server is 4.1.3.0.

We split the storage pools so we would not have to collocate the pool. We
did not want to collocate the pool because we want to get multiple tapes
running at the same time to speed up the backup. It worked.

Werner Kliewer
Manitoba Public Insurance