Thomas,

I restore files like that on daily basis w/o problems. Yesterday I even
restore files from hp-ux 9.04 (now dead! RIP) and a hp-ux 10.20 on VxFS.
What's your patch situation against VxFS?

Mauro/STM

> I have been attempting to restore selected files from a production
system to a
> test system. I start the GUI client with the -virtualnodename option,
select a
> mount point directory, de-select some of its subdirectories, and
request that
> the selected files be restored to a different mount point directory on
the
> test system. Both systems are HP-UX 10.20 and both have the 3.1.0.6
client
> installed. The /etc/mnttab files on the two systems indicate that both
the
> original and the target file systems are VXFS file systems. The GUI
will
> restore some files and then display a dialog box reporting and internal
error
> and referring the user to dsmerror.log for further information. A
typical
> dsmerror.log is shown below:
>
> 03/07/00   15:11:50 TransErrno: Unexpected error from open, errno = 22
> 03/07/00   15:11:54 Internal system error
> Please check the ADSM Error Log for any additional information
>
> In some cases, the error is from mkdir rather than open. In cases where
I have
> run more than one test in a day, all the failures resulted in identical
> messages, with exactly the same number of files successfully restored
before
> the failure occurs. Tests run on different days sometimes result in
slightly
> different messages, as noted above, and consistently restore different
numbers
> of files before failing. The differences from day to day presumably
result
> from running incremental backups of the production server. One of our
systems
> administrators recreated the target file system on different physical
disk
> space because he suspected disk hardware problems. This change had no
effect
> on the results of restores. I wondered about corruption of the ADSM
software,
> and ran Unix sum commands against the dsm and dsmtca binaries on the
test
> system and on one of our production systems. The checksums matched
exactly.
> Does anyone have any further ideas for getting to the bottom of this
problem?