>I'm trying to use HSM managed file system mounted with nfs on Digital unix.
>When I try to work with files through nfs, which are migrated - for example
>cp xyz /dev/null, I'm getting error messages like I/O error - readed
>incrorect number of bytes, or Device busy. This messages I'm getting
>immediately after recall proces has finished. When I run the same command
>again, on the second or third time is it OK. Does anyone know where the
>problem could be? Could hsm file systems be mounted throug nfs?
>Tvoli HSM client is v.3.7.2 installed on Solaris 2.6, VxFS v. 3.3.2.
>Mounted through nfs on Digital unix v. 4.0f ( problem is, that HSM client
>does not exist on Digital unix).

Viliam - It is legitimate to use HSM through NFS.  What is supposed to
         happen is for the nfsd on the NFS server to be caused to wait
for the recall to complete and then respond to the NFS client with the
data.  Apparently, HSM on your NFS server system is not causing the nfsd
to wait: instead, it is returning to the NFS client an indication that
the data is wacky and unavailable.  NFS client caching retains this
state information for a certain amount of time, after which it times
out, and your repeated access attempt finally succeeds.
    One thing to check is if your NFS client system is hard-mounting
the HSM file system (vs. soft-mounting) so that operations on that
file system are forced to wait for them to complete.  Beyond that,
you'll have to confer with your Digital OS folks for examination of
the problem.
         Richard Sims, BU