It really sounds like you're simply overwhelming the broker with those 40,000 messages you're publishing. If the broker couldn't keep up with the volume of messages you were sending (or worse, if you ran it into a limit and the broker stopped accepting messages until something started consuming them), that would be consistent with the behavior you're seeing. When this is happening, is the broker or any of its destinations running into any limits (including available space on disk if applicable)? I know you said "broker resources are within limit", but does that apply to the queue to which you're publishing as well? Is the broker burning all the CPU cores on the machine on which it's running? Is the broker responsive to activity on other destinations and via web console and JMX?
The fact that you lost messages while using async sends seems like the same fundamental problem: the broker can't keep up with what you're asking it to do, for one reason or another. If you switch to sync sends (just for testing) and enable Producer Flow Control, does the behavior change? On Thu, Oct 15, 2015 at 7:13 AM, dhananjay.patkar < dhananjay.pat...@gmail.com> wrote: > Yes all the messages are getting published to the same broker. > > Producer window size was introduced to avoid message loss, previously we > used to lose ~10-20% of messages as we use async send. Usually we get huge > number of messages in short time (message burst). > > > > -- > View this message in context: > http://activemq.2283324.n4.nabble.com/ActiveMQ-consumer-count-is-set-to-0-tp4675930p4702994.html > Sent from the ActiveMQ - User mailing list archive at Nabble.com. >
