Something even simpler:

If the scheduler service is dying (as opposed to hanging), just change
the Windows properties of the scheduler service (on the properties
RECOVERY tab) so that if the service dies, Windows restarts it once
automatically.  I found that to be quite effective for Windows at the
5.1 and 5.2 client, where the scheduler would just keel over about once
a month per client, on average.  That strategy works fine, avoids a lot
of the missed schedules.

And FWIW, in my experience, the scheduler process dies much less often
at 5.3.2 and above  (fewer memory leaks than in 5.1 and 5.2).  
And you need to get to 5.3 anyway - 5.1 is already unsupported, and 5.2
support goes away as of Mar. 31.  And you should be running 5.3 anyway
to get good System State backups of your Win2K3 machines.

So if I had to choose where to put effort, I would focus first on
changing the Recovery properties, and getting the client upgraded, as
being more important than converting to prompted and implementing CAD.

Maybe do those things first, then convert first to CAD any machines that
run multiple schedulers or are repeat offenders.
But YMMV.

Wanda Prather
"I/O, I/O, It's all about I/O"  -(me)

  
 

-----Original Message-----
From: ADSM: Dist Stor Manager [mailto:ADSM-L AT VM.MARIST DOT EDU] On Behalf Of
Schneider, John
Sent: Wednesday, January 10, 2007 1:49 AM
To: ADSM-L AT VM.MARIST DOT EDU
Subject: Does Client Acceptor Daemon improve the reliability of Windows
ba ckups

Greetings,
        I recently became responsible for an environment with over 700
Windows clients, spread out over 11 TSM servers.  I am on a mission to
improve the reliability, since on any given day there are 10-15 clients
that
don't backup, mostly because their TSM scheduler process dies.  The
first
thing I am doing is working with the Windows team to upgrade to the
latest
client code (some of the TSM clients are running 5.1 and 5.2!).  Another
thing is to switch them all to "schedmode prompted", since most of them
are
still polling.
        But another thing I am thinking about changing is to stop using
the
traditional TSM scheduler process that runs all the time, and
configuring
'managedservices webclient schedule' and the Client Acceptor Daemon.  My
habit has always been to use CAD, and this is the largest shop I have
seen
that doesn't use it at all.  It has always seemed to me that the CAD
process
would make things more reliable, since the CAD process wouldn't die even
if
the scheduler process did, and the CAD process could launch the next
backup
without intervention on someone's part.
        But I would like to hear from you with a lot of experience in
large
shops.  Does the CAD service really make TSM client backups more
reliable?
Is it worth the effort to convert to?  I don't want to heap more work on
the
Windows staff unless there is a payoff.

Best Regards,

John D. Schneider
Sr. System Administrator - Storage
Sisters of Mercy Health System
3637 South Geyer Road
St. Louis, MO.  63127
Email:  schnjd AT stlo.mercy DOT net
Office: 314-364-3150, Cell:  314-486-2359