I have a service “getOrders” that connects to a database and grabs some orders and processes them for me. It can take anywhere from a minute or two to ten minutes. I need it to run serially, so I set up in scheduler a repeating task with a simple interval. And I gave it an interval of 60 seconds, and I checked the “Repeat from end of invocation” box. I’m running 6.0.1. It works great for a week or two, and then what I find is that it simply hangs. You can tell that it thinks its still running because it says the next run is in “9223370965949572.0 sec”. However, it never changes from that, it just continually thinks that the service is still running. I’m stumped. Since it works for a week or two before this occurs, I was wondering if it was a known bug in scheduler or something. I can run the service manually, and it works fine. The only way I can get it to work in scheduler once it’s stuck is to restart WM, but that’s not something I’d like to have to do in production. Any ideas?
Jamie,
We ran into Scheduler issues ourselves. It sounds like similar situations. We are running a Unix platform with JVM 1.3.1, however we had to back off one version from 1.3.1_09 to 1.3.1_07. This corrected our problem. I’m not sure if wM is looking into correcting this or not.
Hope that helps.
J.J.
Sadly, webMethods has a fundamentally flawed scheduler, the primary flaw being that it hangs. You will find other problems, like the fact the scheduled jobs share a single session (a potential disaster when dealing with database connections). We’ve had to write silly scheduled jobs which write to a file, then have some external Unix process watch for that file so we can detect scheduler hangs. These hangs can bring your server to its knees, since the various webMethods housekeeping services (i.e. session expiration) use the scheduler.
I rather dismayed to hear you report the problem still exists in 6.0. However, given my experiences with Developer 6, in which many things that once worked are now broken, I’m not suprised.