>From the sound of it, no one queue is overloading the broker; rather, you have lots of queues that each have some messages, and the aggregate load is the problem. So don't look for the problem child, and instead focus on protecting the entire broker.
First, you should enable Producer Flow Control, since its sole purpose is to prevent the broker from running out of resources due to the receipt of new messages before they put the whole broker in a bad state. PFC stops accepting new messages when a limit you set is reached, and doesn't accept new ones till existing messages are consumed and resources are freed up. Next, you should evaluate whether the heap size you're currently using is large enough to support the load you expect to have. If you're running out of memory and your heap is tiny, it might just mean that whoever handed this configuration to you didn't size it appropriately. 1GB should be able to handle a pretty decent volume and would probably be big enough for the setup you've described as long as your messages are a reasonable size, and certainly 2GB should be plenty of room for most installations, so if you get that high and are still running out of memory, something else is going on. Also, what GC strategy are you using with your JVM? And how do you know you're OOMing? JMX doesn't require an installation, just configure it in your config file. Set useJmx="true", like <broker useJmx="true" brokerName="BROKER1"> See http://activemq.apache.org/jmx.html for more details. You should enable JMX; it's the most useful tool for troubleshooting an ActiveMQ broker, so everyone should have it enabled (but secured with a username and password). Tim On Apr 14, 2016 8:35 AM, "thompsg4416" <grahamthomp...@gmail.com> wrote: Thanks for the reply Jim. We do have the webconsole enabled and I am looking at the queues page. These are all fairly active queues - I.E there are currently about 20 queues with between 20-80 messages in them. The problem is and what I've read is that the messages are no longer dequeing. My limited experience and from what I've read, once you're beyond your out of memory all messages stop dequeuing. I restarted the service but it reloaded the previous messages from the datastore. The only way I've figured out how to get around this is to delete the DB and log files under the kahadb folder. I was hoping the activemq.log would give me more information I.E what queue is overloading the system but it's not. Unfortunately we do not have JMX installed. any other ideas? what can I do? -- View this message in context: http://activemq.2283324.n4.nabble.com/Stranded-ActiveMQ-constantly-Crashing-tp4710773p4710775.html Sent from the ActiveMQ - User mailing list archive at Nabble.com.
