Hello,

I'm new here. I evaluated Active MQ in the last days. Therefore I also made
a performance test. This mostly worked fine. But round about 3 times I had a
big performance issue:

One of my test was a ping-pong. The producer application creates a
"ping-pong-10000" message. This was received by the consumer application and
answered by "ping-pong-9999" which was again received and answered by the
first application until the counter reached 0. So I checked the performance
with a bidirectional communication. This test worked mostly fine. But round
about 3 times (from round about 100 tests) this test take ages. With
debugging/println messages I have seen that one client took 30 seconds for
each response. One time even both clients had this delay. For me this looks
like a timeout because each message took exactly 30 seconds. But no error
message was posted on client or server. Two times I had to restart the
server and all worked again. One time even this did not worked and I needed
to remove the persistence of ActiveMQ to go back to normal production.


Environment:
OS: Windows XP
Active MQ: 5.4.2 (out of the box, so with kahadb)
Clients: Java, sometimes also one client was C++ (using ActiveMQ-CPP-3.2.4)
Settings: Communication was always done with a standalone server, using TCP
and openwire.
I'm not sure for all cases but the last time it was with persistence for
sure! I belief that it was most often with persistence and probably queue
communication (so the slowest of all options).


Can anybody explain the problem or help me how to avoid this issue?

Thanks,
Martin
-- 
View this message in context: 
http://activemq.2283324.n4.nabble.com/Bad-performance-timeout-sometimes-tp3324131p3324131.html
Sent from the ActiveMQ - User mailing list archive at Nabble.com.

Reply via email to