server-side stomp timeout?

[Christopher Wood <ch...@pobox.com>]

My ActiveMQ daemon is not disconnecting dead stomp consumers. Is there a server-side timeout setting that I can use somewhere? I'm having difficulty finding one.

I have a chaos monkey intermittently killing processes on my consumer hosts, and if this happens when the consumer is slurping down messages, the ActiveMQ host will keep the messages in its queue and never let another consumer take them. (By "never" I mean three hours and counting now.) Restarting ActiveMQ clears the issue.

Another puzzling item is that ActiveMQ now reports that I have more consumers connected than I know are trying to connect. I suspect some are the now-dead consumers but I'm not sure how to confirm. However, they are from the same hosts as previously.

I use Net::STOMP::Client with this uri (munged hostname):  stomp://mqlab:61612

Things are fine until one of the consumers dies while slurping down messages, and I am quite stumped.