Hi I am using AMQ 4.0.2 RC6 and also Servicemix 3.0 incubating. I have the Amq instance contained in Smix bridged with the 4.0.2 instance. I am doing some testing, hoping to elicit some tweaks that will shore up the message flows.
So here is what I tried recently: 4 STOMP ruby producers sending to the Smix Amq instance, Smix routes the messages through content based router, then pushes them onto a queue. A Windows service uses the .NET OpenWire api to consume messages from this queue. It then pushes the consumed messages to a enterprise app. Both Amq instances are using separate MySQL databases to persist messages. Here is the test: each of the 4 STOMP clients produce 500,000 messages to queue in Smix. Smix has handled them just fine, passing them to 4.0.2 Amq instance via the bridge. After some time the 4.0.2 ran out of heap! Kind of a slow consumer deal, but I would have figured that the heap would go down as the messages persisted to the database. There are 892281 messages in our MySQL DB, post 4.0.2 crash. Is there an issue with messages not being flushed from memory? I would think only the persistent messages needed by the journal would be in current memory. The rate at which the message flowed into the 4.0.2 instance was about 10 messages per second. Could it be the messages were not being flushed to the DB quickly enough? Would I need to set this to happen in a shorter interval? Oh also something I noticed was the Smix Amq instance did not persist any messages, they all seem to be in the journal, about 600K worth. I'll see if I botched the config or something. thanks, Rob Ottaway -- View this message in context: http://www.nabble.com/Enterprise-messaging-set-up-needss-tweaking-tf2608885.html#a7280417 Sent from the ActiveMQ - User mailing list archive at Nabble.com.
