1. Our upgrade experience from 5.3 to 5.5 was a little while ago, and it
mostly went smoothly, except that the promised automatic UPGRADEDB
didn't happen. But this was obvious upon first trying to start it, and
it was simple to do it manually. Clearly nothing more than an
inconvenience.

2. Don't stop at 5.5.0 - we had some serious crashes with it which have
been fixed by 5.5.1. If reclamation starts while there is a Restartable
Restore pending, 5.5.0 will crash. It will also crash if a client node
starts a restore for files that reside on volumes currently involved in
reclamation. See APAR IC55710, which is fixed in 5.5.1. We have verified
that 5.5.1 fixes this problem.

When this happened to us, we were left in an impossible situation. The
storage pools were set to start reclamation (that is, there were volumes
in a storage pool with more reclaimable space than the current threshold
setting), but there was no way for us to enter a command to reset the
threshold to keep reclamation from starting, or to cancel the
restartable restore, because every time we brought it up, it was already
crashed by the time we tried to enter a command. The bypass was to place
the following line in our dsmserv.opt file:

  NOMIGRRECL

which is an undocumented server option that prevents migration and
reclamation from starting. Then we were able to start the server, cancel
the restartable restore, set the thresholds for no reclamation, upgrade
to 5.5.1, and so on back to recovery.

Roger Deschner      University of Illinois at Chicago     rogerd AT uic DOT edu
               Academic Computing & Communications Center
====You will take a long journey. Remember to export your variables.====


On Mon, 25 Aug 2008, Lepre, James wrote:

>  We are currently at TSM v5.4.1.1 and are planning on upgrading to
>v5.5.1.  Can anyone let me know if there are any caveats that I should
>be aware of, or any new features that people like in this version.
>
>Thank you
>
>James Lepre