Hello Jean, It appears that you are using the C++ broker and the java client. Could you confirm which version you are using?
On Tue, Jan 5, 2010 at 12:47 PM, Jen Andre <[email protected]> wrote: > > Hello, > > I have a Java client listener that subscribes to a topic and waits for > messages. I'm noticing after 120 seconds of idle time (no messages > received) the connections time out (here's what I see in syslog messages for > qpidd) > > qpidd[3472]: 2010-01-01 16:46:20 error Connection timed out: closing > > This is consistent with the debug messages I'm seeing on the client side: > > IoReceiver - /192.168.10.128:5672 2010-01-05 12:32:10,126 WARN > [apache.qpid.transport.ClientDelegate] Ignoring the idle timeout 0 set by > the connection, using the brokers max value 120 > > and > > > IoReceiver - /192.168.10.128:5672 2010-01-05 12:38:57,898 DEBUG > [apache.qpid.transport.Connection] RECV: [conn:9ced8e] ch=0 > ConnectionHeartbeat() > IoReceiver - /192.168.10.128:5672 2010-01-05 12:40:57,899 DEBUG > [apache.qpid.transport.Connection] RECV: [conn:9ced8e] ch=0 > ConnectionHeartbeat() > IoReceiver - /192.168.10.128:5672 2010-01-05 12:40:58,618 DEBUG > [apache.qpid.transport.Connection] connection closed: conn:9ced8e > > > I'm a bit confused on how timeouts and heartbeats are supposed to work. > Does the client force the disconnect, or the server, or both, if the > heartbeats are not received in a specific period of time? Both the client and the broker could enforce a disconnect if they don't see any activity during two consecutive heartbeat intervals. Do both peers > send heartbeats? (this seems to imply it does. Both peers do send a heartbeat. The JMS client (0-10 version which you are using against the c++ broker) will echo a heartbeat in response to one sent by the broker. > http://qpid.apache.org/configure-broker-and-client-heartbeating.html). The link you provided is for the JMS Client (0-8 version) and the Java Broker. At the moment it's slightly different from JMS Client (0-10) and c++ broker combination. The amqj.heartbeat.delay is only effective for the 0-8 codepath. If you need to set the idle timeout for the 0-10 client you need to set it in the broker URL as follows amqp:///test?brokerlist='tcp://anotherhost:5684?idle_timeout='60'' Please note the idle_timeout is in secs and not in milisecs. We are working on consolidating the configuration and behaviour of the JMS client for both 0-8/9 and 0-10 versions. > I do notice that while I see a lot of RECV: heartbeat debug messages on the > client side, I don't see any SEND: messages that indicates it's sending them > to server. If the client is supposed to send them to the server > periodically to ensure the connection stays alive, how can I ensure this is > happening? (setting system property amqj.heartbeat.delay seems to have no > effect). > > Thanks! > Jen > > > > > > -- > View this message in context: > http://n2.nabble.com/timeouts-tp4256406p4256406.html > Sent from the Apache Qpid users mailing list archive at Nabble.com. > > --------------------------------------------------------------------- > Apache Qpid - AMQP Messaging Implementation > Project: http://qpid.apache.org > Use/Interact: mailto:[email protected] > > -- Regards, Rajith Attapattu Red Hat http://rajith.2rlabs.com/ --------------------------------------------------------------------- Apache Qpid - AMQP Messaging Implementation Project: http://qpid.apache.org Use/Interact: mailto:[email protected]
