Zoltan Forray/AC/VCU wrote:

We are running 4-servers. [email protected], [email protected], [email protected]

The 5.5 servers have given us grief. We have at least 4-known (apars)
problems we are waiting on patches for. 1-requires occasional restarts of
the server (this is the IBM workaround).  The other one has crashed
frequently.  In fact, we currently have an open problem on this server.
IBM had to  create us a "diagnostic kernel" version of dsmserv to try to
reproduce the problem and get additional information.

The biggest has a 160GB DB, which is showing us the most problems. Expires
are running 48-hours+.  We only backup 1-2TB daily on this server but have
an occupancy of 177TB with 289M files which I think is where our drag on
the expires, is.

I will not upgrade the non-5.5 servers until IBM gets these problems
resolved.  5.4.2 is the highest I recommend!

Out of curiosity, what hardware and OS are you running? We've had a
couple segfaults in 5.5 when there were a bunch of transactions running
during a database backup, but none since I upped the LOGPOOLSIZE to 2048.

Even with the segfaults, I don't think we'd be able to go back from 5.5
because 5.5 solved a long-standing bug with SET EVENTs on archives.
Whenever we tried to activate retention on an archive with more than a
few hundred thousand files (read: all of our archives) we'd get an
aborted transaction on the client.

--
-- Skylar Thompson (skylar2 AT u.washington DOT edu)
-- Genome Sciences Department, System Administrator
-- Foege Building S048, (206)-685-7354
-- University of Washington School of Medicine