I think we had a thread open for the same issue last week. What you could 
probably do is to add a BAT or CMD file at the root  folder and schedule a task 
to verify the status of the service and start it if needed.

Here is the Bat file I've created using resource kit tools and it works without 
any issues and you don't have to start the service manually on each failing 
server.

Sample BAT File
You need netsvc.exe from Resource Kit.

Actual Code
@echo off
netsvc "Service Name" \\<server name> /query
if errorlevel 0 netsvc "service name" \\<servername> /start
you could also add another line to display a message;
if errorlevel 1 echo TSM Client Scheduler running
exit

Once you have this bat file copied on all the servers being backed up, schedule 
a task on all the servers using "at" command
at 06:00 /every: M T W Th F S Su c:\verify.bat

This eliminates any need to logon and manually start the service.

Hope this helps,

Prasanna
EDS
--

On Fri, 27 Sep 2002 09:29:15
 Rushforth, Tim wrote:
>APAR IC34619 has been opened for this:
>
>APAR= IC34619  SER=                            IN INCORROUT
>WINDOWS SCHEDULER SERVICE FAILS ON STARTUP WITH
>EVENT LOG ENTRIES 7022 AND/OR 7031.
>
>Status: OPEN                                Closed:
>
>Apar Information:
>
>RCOMP= 5698ISMCL    TSM CLIENT 5.1  RREL= R51W
>FCOMP=                              PFREL= F     TREL= T
>
>
>Return Codes:
>
>Applicable Component Level/SU:
>
>
>Error Description:
>Up reboot of a client machine, the TSM scheduler service
>
>It has done this 1 time(s).
>The following corrective action will be taken in 0 milliseconds
>.
>It has been seen that you may only see the 7022 in the logs and
>still witness this apar.
>.
>In a registry trace of the scheduler service, you can see a
>large time gap from when we request the password from the
>registry, and when we actually get it passed back.
>.
>5/16 12:51:19 dsmcsvc.c(1006): GetRegistryEntries(): Opened
>RegistryKey 'SYSTEM\CurrentControlSet\Services\Backup
>\Parameters'
>5/16 12:51:19 dsmcsvc.c(1087): GetRegistryEntries(): Locating
>Registry Entires for Node 'NODENAME' ...
>!!! timeout 2:06 !!!
>5/16 12:53:25 dsmcsvc.c(1171): GetRegistryEntries(): Found
>Registry Password for node 'NODENAME'.
>.
>Service trace:
>
>05/16/2002 12:51:20.0042 [79] : session.cpp
>(1837): sessInit: Session initialized.
>05/16/2002 12:51:20.0042 [79] : session.cpp
> (1919): sessOpen: Attempting to open communications
>05/16/2002 12:51:20.0042 [79] : commtcp.cpp
>(1179): TcpOpen: Internet socket defined.
>05/16/2002 12:51:20.0042 [79] : commtcp.cpp
>(1298): TcpOpen: Socket window size set to 64512 bytes.
>!!!!time out 2:04 !!!!!
>05/16/2002 12:53:24.0820 [79] : pscomtcp.cpp
>(1175): Attempt connection was rc = 0.
>.
>Although customer receive this error on startup, they are
>able to start there scheduler service manually after receiving
>these errors.
>.
>This problem has also been witnessed at 4.2.2.x and
>5.1.x.x as well.
>.
>
>Local Fix:
>Start your scheduler service manually
>
>
>
>-----Original Message-----
>From: Eliza Lau [mailto:lau AT VTCAT.CC.VT DOT EDU]
>Sent: August 22, 2002 2:52 PM
>To: ADSM-L AT VM.MARIST DOT EDU
>Subject: Re: 5.1.1 Windows 2000 scheduler won't start
>
>Tim,
>
>I haven't heard back from Level 2.  I don't even know what APAR it is given.
>The scheduler fails to start the first time after we installed the 5.1.1
>client on a Windows 2000 server (not Professional).  The event number in the
>event log is
>7031  The TSM scheduler service terminated unexpectedly
>
>After the machine reboots, the scheduler doesn't start even though it is set
>to auto.  The event number is
>7022  The TSM scheduler service hung on starting
>
>Starting it manually then after works.
>
>We can recreate it every time the machine boots.  This is a new Windows box
>with a new w2k server installed on it.
>
>Eliza
>


____________________________________________________________
Tired of all the SPAM in your inbox? Switch to LYCOS MAIL PLUS
http://www.mail.lycos.com/brandPage.shtml?pageId=plus