Starting Software AG Universal Messagin Service

I start my webMethods server every 3 weeks. My server have the following service

  • CentraSite registry repository
  • Integration Server
  • my webMethods Service
  • Optimize Analytic Engine
  • Optimize Web service data collecter
  • Platform Manager
  • Runtime
  • System Management Hub
  • Universal Messaging

The problem is when the server start the week end, the Universal Messaging service doesn’t start (even if I put it as Automatic deffered type of start). I think it’s do tue the the service need another service to start. So I want to set some dependences to Universal Messaging service but I don’t know which service need to be start.

to find out, you can shut everything else, and try to bring up UM manually. There should be some error messages generated, so you can tell which one.


you can check in the event log of Windows for messages regarding UM service.

There should be a message why UM did not start.

Most likely it is due to a file system not yet mounted which UM needs access to.


Also remember to check your UM log files (e.g. UniversalMessaging\server\umserver\bin\UMRealmService.log and UniversalMessaging\server\umserver\data\nirvana.log). If you don’t see anything there, try starting UM from the command-line rather than as a Windows service. If it doesn’t start, then you know your problem is not related to how the Windows service is configured.

By the way, you shouldn’t have to set any dependencies for UM. In fact, UM is typically the service I start first.

Last but not least, why are you restarting everything every 3 weeks? Sounds like there’s an underlying problem here.


I don’t decide to restart server every 3 weeks, it is scheduled by our network service to be sure than fix and update from Microsoft are up to date in each server (not especially SAG server).
Beside it is VM machine (the physicial server have other VM in the server and need to be restart).
So the server need to be restart.

I will plan to start um first. It make sense for me because some problem can come from the fact that the UM server is start, however connection JMS to UM doesn"t work when server restart

Got it. Just wanted to make sure you guys weren’t restarting due to a memory leak or something like that.

In my experience, UM starts so much faster than the IS that if you have them both set to Automatic, then by the time the IS finishes starting, UM is already up so no connection issues occur.

From your list of products, the only product I recall setting to Automatic (Delayed) was Optimize’s Analytic Engine because there are known issues with AE hanging on a server startup when the Windows service is set to Automatic (e.g.

Another thing you can consider, if your license allows it, is to break up some of these components into separate VM’s. The fact that all these things are attempting to start at the same time could be causing slow down’s and timeout’s.


In fact,
IS and UM are started automatically but sometimes I found that UM does’nt start and I dont know why.
So I guessed it was du to the fact that UM need some other services to start but maybe it’s the opposite :UM doesn"t wort for a reason or another if it is not the first service start.

Other problem is the problem of connection : sometimes when I see the JMS connections in IS after restart of the server, I see error (in red) in IS one some JMS connection like that the bus is not enables (even if the JMS connection is not disabled).

So I’m looking for the reason of this problem.

And to be honest with you : actually we have only one IS and UM server but for reason of high availability we need more and we’re looking for partners to help us to evaluate the licence cost to upgrade it and migrate solutions

After setting a dependency to IS as UM need to be start first, I think I change order of starting services. So it’s not the same order right now.
Now it is CentraSite which not start. So I guess the problem is there is too many services which start in same time and windows doesn’t succeed to restart all service.

It is possible ?

It’s possible. It would require some digging through logs, including Windows event log, to see if we can find specific errors to confirm the theory. But yes, if we have too many things trying to start at once, it is possible that the CPU is getting pegged, slowing the start up of those services down, and causing timeouts. Products that rely on the Tanuki wrapper can be especially sensitive to timeouts because the Tanuki wrapper has a built-in mechanism to ping the JVM, and if the JVM is deemed unresponsive, the wrapper will take action:


OK I solve the issue by restart the service after the first failure in recovery tab of service (sorry I translate it from French to English but I m not sure it is the term used in windows specification in English) for each WM service and it seems to work.
Thanks to help me