We have a production environment in which there are several remote servers which will go online next week (10 servers). We are using JMS queues to store messages to be sent to the servers when they go 'online' for updates. One server for example has over 34,000 messages in it's queue ready to be processed. In the meantime we have 3 servers with their own queues already in production. Everything was running fine for the past 2 months, until over the weekend when at 3am the JBOSS server was 'hung' and we had to restart it in the morning. At 3am we have nightly updates to be sent to the enterprise server. For some reason the nightly updates did not appear on the enterprise server. We had to resolve this issue manually. However, when we restarted the JBOSS server the message processing is again running smoothly as it has for the last 2 months. We do notice a performance degregation later in the day.
What should we look at to resolve this issue? Keep in mind that we will have 180+ servers and 180+ queues when we are fully 100% in production.
The enterprise "JBOSS" server is a Pentium IV computer running Windows 2000 Server Edition with 2 gigs of memory.
We changed the hsqldb-ds.xml setting for max pool size from 20 to 100 today to see if this improves the performance.