Thanks for the link with regards to keepAlive and the configuration of soTimeout. Will experiment with these settings in 4.1 for now. Am a bit hesitant to deploy 4.2 on our system as it is not considered a stable release as of yet. Though, I will checkout your 4.2 and validate that this issue no longer does occur.
It does seem a bit interesting that the two exceptions (SocketTimeoutException and InterruptedIOException) are caught but not handled within TcpTransport. Was there a reason for this? tvijlbrief wrote: > > One last note on the subject of crashed clients, the current 4.2 SVN > version > works in my environment (Linux, JDK 1.5) just fine out of the box. > > If a client is suspended on Linux with CTRL-Z > then the broker closes the connection after 30 seconds. > > I'm not sure why eccrowe has the long delay problem: > > ===quote:==== > Using the jmx-console I see that clients are eventually removed > upon power outage; but, it is taking approximately 15 minutes for that to > occur. > =========== > > This 15 minutes looks like the Linux default soKeepAlive timeout: > > ========== > An idle > connection is terminated after approximately an additional 11 minutes (9 > probes an interval of 75 seconds apart) when keep-alive is enabled. > =========== > > I don't understand why on 4.1 (with the patch of the InactivityMonitor) > the > broker does not close the connection > after 30 seconds, has this code changed in 4.2? > > It is also not clear why the default delay of 7200 seconds before > the soKeeplive kicks in, is not > executed although soKeepAlive is not set on the socket by default. > > I would have expected that eccrowe would have had to wait over 2 hours. > > Tom > > > > -- View this message in context: http://www.nabble.com/Memory-usage-increase-after-client-power-failure-tf3242706s2354.html#a9162399 Sent from the ActiveMQ - User mailing list archive at Nabble.com.
