=================================
I. DOWNLOAD INSTRUCTIONS
=================================
1) Download J0850483.<platform>.tar into your /tmp directory.

    where <platform> is hp_11 hp_dm sgi_dm solaris solaris_dm
 
    NOTE: Solaris7 machines use the "solaris" named platforms

2) untar J0850483.<platform>.tar
       /bin/tar xvf J0850483.<server>.tar

   This will create the files:
   Vrts_patch.install
    patchJ0850483.README
   patchJ0850483.<platform>.tar.Z

3) Read patchJ0850483.README for complete details
  of the problems fixed with this patch and
  installation instructions.

==================================
II. DESCRIPTIONS OF PROBLEMS FIXED
==================================

=============
current patch
=============

Note: This patch contains updates to the following man pages;
     migmode, migmdclean, migrc and (new) migactivate  

Description:
  1) HSM needs a way to activate FHDB entries.
     If a file is backed up and then restored to an alternate path
     there will be two files in the filesystem that have the same HSM handle.
     If one of these files is deleted the FHDB entry for the handle will be
     marked obsolete. If the volume that the files copy resides on is
     migmdcleaned the file that was not deleted can no longer be cached.
   
   2) FLUSH feature exceeds the limit of files to write before writing a
      tape mark.  If "migcopy" reaches end of tape the continuation tape will
      copy the wrong number of files to the continuation volume before the
      next tape mark is written.
   
   3) "migtrans" must not continue if "mt_newdestvol" fails.
     If, during consolidation, "migtrans" is executed with the "-f"option "migtrans"
     ignores all errors returned when selecting a new destination volume.
     This can cause "migtrans" to look for destination volumes forever.

Workaround:
    1) "migsetdb" can be used to activate FHDB entries.
   
   2) There is no workaround, this problem does not cause any data loss,
     the files still can be cached from the continuation tape.
   
   3) Do not alter the migcons script to pass a "-f" to "migtrans"

Additional Notes:
     "migrc" has a new option "-S" that causes "migrc" to execute "migactivate" to
     activate all FHDB entries (not dk) for all files in the managed filesystem
     that have HSM handles with FHDB entries that are not active.
     NOTE: The -S option must be used with another option such as :
            -O FHDB , -O VOLDB, etc.
   
     "migmdclean" has a new option "-S" that works the same as the "migrc -S" option.
   
     "migloc" has been changed to show volumes even if the FHDB entry is obsolete.
   
     "migactivate" is a new HSM command that is executed by "migrc" and "migmdclean"
     if the "-S" option is specified.   "migactivate" may be executed by the administrator
     instead of running "migrc" or "migmdclean".
   
       Usage:
         migactivate [-v] hsm1
           -v will cause all messages to also go to STDOUT or STDERR.
           If the HSM log level is set to 8, migactivate wil list all HSM
           handles that have been activated and each files inode number.

      (All UNIX Storage Migrator servers)

Description:
   (1) In the 3.4 release, code was added to pre-allocate space in a VxFS
        file system when caching back a file.  The pre-allocation code was
        often not allocating enough space.
   
   (2) When two "migcopy" processes are making (outward) copies of the same
        file, the one that completes first will wait until the second "migcopy"
        gets its volume mounted.  This was because "migcopy" was holding a DMAPI
        shared lock on the file, preventing the other "migcopy" from briefly
        obtaining an exclusive lock.  Fixed by not holding a lock during the
        mount process.
   
   (3) In a VxFS file system, a user initiated "SIGSTOP" (ctrl-Z from the
        controlling tty) could prevent a "migcopy" process from proceeding
        while the user process was stopped.  This problem was corrected by
        using a new DMAPI token for the "migcopy DMAPI" requests instead of
        using the token generated by the DMAPI event.

      (All UNIX Storage Migrator servers)

Description:
     Files migrated to nb volumes when HSM was running for UFS filesystems
     are restored on VXFS filesystem as if they are owned by root.
     This is only a problem for sites that have switch for UFS to VXFS
     and were using the nb method.
     This patch will restore the files with the correct owner and permissions.

      (All UNIX Storage Migrator servers)