Documents get struck in Trigger queue

Hi All,
We are facing a problem with publishable documents. We have 2 pipelines Synchronous which uses publish and wait service ,other Asynchronous uses publish service. Suddenly documents got pileup in Asynchronous Triggers queue. While Synchronous queue was going smoothly.we can see the documents in the queue but they are not picked by client (subscribing service). IS looks ok.As it was a time for IS restart maintainense schedule, after the IS restart documents got processed and cleared. After few days, it happened to Synchronous Queue, and we have not clue why documents are getting pileup in the queue, having no choice we had restart the IS. This is happening in the production environment. Does any one has faced this kind of issue. If so is there any solution other than restarting the IS.what exactly is the issue.

IS-6.5-servicepack 3
Broker 6.1 nopatches
Windows 2000.

hmmm can you see anything on IS server or borker log?

Hi devexpert,
I have checked the server and the broker log and there are
no error message or any kind
of other message

For the trigger configurations, are they concurrent or serial? If serial, it is possible that the trigger service processing the documents is hung. If this happens again, look Service Usage page in Administrator to see if the service is running. What type of work are the trigger services doing?

Hi reamon,
The triggers are Concurrent. I have another question, which I think is related. When I went through the IS error log I found this error
[ISS.0098.0084E] Exception:unable to create new native thread during Trigger dispatching…
Recently we have increased the JVM memory on IS from 1gb to 1.5 gb as it was not allowing 2gb which is the max for jvm on windows system. Does this has to to anything with the document struck in the clients on the broker

Yes, it would definitely have an impact. How many concurrent threads is each trigger configured to use? You probably need to tweak the number of allowed threads, perhaps at the IS and trigger level.

H Rameon,
Currently we have 20 concurrent threads one Trigger and 4 for another one.
Queue capatity : 50 each
Refill level : 15
Document retrieval
Maximum Threads: 300 (100% of Server Thread Pool)
Queue Capacity Throttle 100%

Document processing
Maximum Threads : 300 (100% of Server Thread Pool)
Execution Threads Throttle 100%

The maximum number of native threads allowed per process is defined by the OS. Depending upon the JVM in use a thread pool of 300 may be too big. Given that the error indicates there were no native threads available, it seems safe to say the number needs to be reduced.