Moving a TSM HSM managed filesystem to a new host machine with different Filesystems

PREDATAR Control23

As the article says, the file systems must be the same. Now you could check out using VxFS for AIX and see if that will work. I would call IBM and ask them if there is a migration path between Solaris and AIX for the HSM data.
 
PREDATAR Control23

In this case, it is a hospital, and an archive of radiology images and scanner, which are still in use (in production) ... So if we lose these images, we can not recreate ...
So we can not really play with these data ... But we can not keep data on a system being phased out.
 
PREDATAR Control23

In this case, it is a hospital, and an archive of radiology images and scanner, which are still in use (in production) ... So if we lose these images, we can not recreate ...
So we can not really play with these data ... But we can not keep data on a system being phased out.

If I would implement a move like this, I would build the new system, copy the data over the new system. 're-index' (meaning rebuild the HSM), run in parallel, slowly move users over to the new system, observe and then decommission the old one.

Easier said than done but I believe that this is the only 'easy' way.
 
PREDATAR Control23

The big problem is the volume: 17 TBytes and more than 2 million files. All files belong to the same user and same group.
And we can't stop HSM more than three or four days, because these files are used by practitioners and radiologists to treat patients.
Imagine the delay to restore 17 Tbytes from LTO5.
 
PREDATAR Control23

The big problem is the volume: 17 TBytes and more than 2 million files. All files belong to the same user and same group.
And we can't stop HSM more than three or four days, because these files are used by practitioners and radiologists to treat patients.
Imagine the delay to restore 17 Tbytes from LTO5.

I understand the size as mentioned earlier. Don't you have copies of the tapes? If you have, you can use these to restore the data on the new AIX server.
 
PREDATAR Control23

Sounds like we are dealing with BLOB files here. Moon-buddy, can he really restore the solaris data from a VxFS file system to an AIX JFS2 file system? I think that I would set up the new AIX box and have AGFA go to the new box and a specific date. That way all new data would start to accumulate on the new system. The old system could stay up and you would have to recall the data and then move it over to the new system. It will take a while, but no one would be down.
 
PREDATAR Control23

In this case, it is a hospital, and an archive of radiology images and scanner, which are still in use (in production) ... So if we lose these images, we can not recreate ...
So we can not really play with these data ... But we can not keep data on a system being phased out.

In my opinion and working in the same sort of environment where we've used HSM for the same type of data... Your best bet here is to involve the RPACS vendor and do an archive to archive migration, you'll have an outlay of funds for the RPACS vendor contract and the new AIX system but this avoids any downtime at all and will allow you to throttle the migration of studies in to the new archive area.

Now if you didn't want to go that route when we switched from AIX JFS to JFS2 files systems we had to restore all our stub files, there's a tech article on IBM's site but you may be able to do this from Solaris to AIX. I know the restore GUI will allow you to restore between these two system.

http://www-01.ibm.com/support/docview.wss?uid=swg21286426
 
PREDATAR Control23

Sounds like we are dealing with BLOB files here. Moon-buddy, can he really restore the solaris data from a VxFS file system to an AIX JFS2 file system? I think that I would set up the new AIX box and have AGFA go to the new box and a specific date. That way all new data would start to accumulate on the new system. The old system could stay up and you would have to recall the data and then move it over to the new system. It will take a while, but no one would be down.

Yes I think it can be moved. These images are digitized so it should be transparent. These should be no different than handling other data.
 
PREDATAR Control23

OK, suppose we lose all stub. In this case, the command "dsmc restore ..." should be able to recreate them: Is it true?
If I generate a node (AIX) , with the same dsm.opt and dsm.sys files on the SUN, then create FS (JFS2), with the same mount point, I should be able to recreate the STUB, without restoring data using the following command:

# dsmc restore "/archive/*" -su=yes -replace=no -restoremigstate=yes

In this case, TSM should restore stub files (but not the data files) on the AIX file system (JFS2), even if the original system was VxFS.

Is that true?
 
PREDATAR Control23

I am not sure about the ARCHIVE you have put in there but the manual says:

To restore stub files rather than backup versions of your files, for example if one or more of your local file systems is damaged or lost, use the backup-archive client restore command with the restoremigstate option. To restore the stubs of HSM managed files with the backup-archive client, the dsmrecalld daemon must be running. Your migrated and premigrated files remain intact on the Tivoli Storage Manager server, and you need only restore the stub files on your local system. However you cannot use the backup-archive client to restore stub files for your migrated files, if they have been backed up before the migration. Instead use the HSM client dsmmigundelete command to recreate stub files for any migrated or premigrated files that are lost.

Here is an example: dsmc restore -restoremigstate=yes -sub=yes "/home/*"
 
Last edited:
Top