All,

I'm in the process of doing an initial backup ('dsmc inc /hsm') of a
large ADSM HSM filesystem (AIX 4.2.1, ADSM 3.1.0.6, the migrated
dataset is large, the disk is not so large).  The ADSM server used for
backup and hsm are the same.  According to the ADSM docs (Chapter 6 of
"Using the UNIX HSM Clients"):

  Backing Up Files After Migrating Them
  -------------------------------------

  If files are assigned a management class that does not include the
  requirement for an existing backup version, you can still back up the
  files after they are migrated.

  If you back up the files to the same ADSM server to which they were
  migrated, ADSM copies the file from the migration destination to the
  backup destination. It does not recall the files back to your local
  file system.

Earlier in the chapter, it likewise says:

  Backing up and migrating files to the same ADSM server has the
  following advantages:

  o If a file has been migrated prior to the backup, it is cloned on
    the server rather than being sent again over the network.

However, this is NOT the behavior I'm seeing.  Rather, all migrated
files are been recalled, resulting in the files being sent over the
network twice (from server to client and back to server again) and the
HSM filesystem filling.

Any ideas as to why this is happening, and how I can make it work as
advertised?

As perhaps a relevant note, the files on the HSM filesystem where
originally migrated to another ADSM server (server A).  However, I
have since exported/imported the node's data (only spacemananged
files) to the backup server (server B), and the client's dsm.sys now
refers only to server B (and dsm.opt is empty).

John

-------------------------------------------------------------------------
John Valdes                        Department of Astronomy & Astrophysics
John Valdes                        Department of Astronomy & Astrophysics
j-valdes AT uchicago DOT edu                               University of Chicago