activemq-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Justin Bertram <jbert...@apache.org>
Subject Re: Apache Artemis - Stress-test time
Date Mon, 10 Apr 2017 15:11:10 GMT
> Can you help us Justin, to understand why ??

Not until you get me something more concrete like a reproducible test-case, output from a
profiler indicating where the hot-spot is, or even a series of thread dumps from when CPU
utilization is high.


Justin

----- Original Message -----
From: "alessandro zannini" <alessandro.zannini@bticino.it>
To: users@activemq.apache.org
Sent: Monday, April 10, 2017 9:56:19 AM
Subject: Re: Apache Artemis - Stress-test time

Hello, 
I'm a colleague of Francesco. 
We just made another test , just connecting, without subscribing, in order
to check if the problem is the authorization procedure... and we discovered
that this step doesn't occupy a lot of cpu (never reached 5% of usage).
So we can say that the "heavy" step is the subscription procedure made after
connection. For each client , we made a subscription to 50 topics (different
for each clients). 
In our test we subscribed 2000 clients (so, 50x2000 = 100K topics).
During this period (that has requested 1h), the CPU of broker during
subscriptions was ever near 100% of usage! 
Can you help us Justin, to understand why ??
Thanks.
  



--
View this message in context: http://activemq.2283324.n4.nabble.com/Apache-Artemis-Stress-test-time-tp4723999p4724785.html
Sent from the ActiveMQ - User mailing list archive at Nabble.com.

Mime
View raw message