Mailing-List: contact dev-help@activemq.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@activemq.apache.org
Date: Wed, 22 Apr 2015 05:33:58 +0000 (UTC)
From: "Arthur Naseef (JIRA)" <jira@apache.org>
To: dev@activemq.apache.org
Message-ID: <JIRA.12776131.1424345925000.71309.1429680838983@Atlassian.JIRA>
In-Reply-To: <JIRA.12776131.1424345925000@Atlassian.JIRA>
References: <JIRA.12776131.1424345925000@Atlassian.JIRA>
 <JIRA.12776131.1424345925245@arcas>
Subject: [jira] [Commented] (AMQ-5598) MQTT+NIO+SSL: Concurrent connections
 lead to 100% CPU usage
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit


    [ https://issues.apache.org/jira/browse/AMQ-5598?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14506456#comment-14506456 ] 

Arthur Naseef commented on AMQ-5598:
------------------------------------

I see - thank you for the clarification.  MQTT and AMQP are still vague in my mind and hence I lump them together - I need to find time to learn them.

> MQTT+NIO+SSL: Concurrent connections lead to 100% CPU usage
> -----------------------------------------------------------
>
>                 Key: AMQ-5598
>                 URL: https://issues.apache.org/jira/browse/AMQ-5598
>             Project: ActiveMQ
>          Issue Type: Bug
>          Components: MQTT
>    Affects Versions: 5.10.0, 5.11.0
>         Environment: Red Hat Enterprise Linux Server release 6.6 (Santiago), Open JDK 64bit 1.7.0.75, 3Gig RAM, 30Gig HDD - 1x ActiveMQ 5.10
> Windows 7 x64, Java 64bit 1.7.0, 8Gig RAM, 1x ActiveMQ 5.10, 1x ActiveMQ 5.11
>            Reporter: Josef Awad
>              Labels: performance, security
>         Attachments: Screen Shot 2015-02-12 at 8.31.58 PM.png
>
>
> Connecting with multiple instances of  mqtt clients via ssl at the same time causes the borker server to run on 100% CPU usage forever. Only restart of ActiveMQ helps. Thread count maximizes during connection attempt.
> Critical: It doesn't matter if the certificate is valid or not, I am able to shut down the broker from one client with 500 concurrent connecting instances. The payload size is not important. The problem is somewhere at the ssl handshake located. Without ssl everything works fine. Single ssl connections work.
> Similar issues (with older versions) described are here:
> - http://activemq.2283324.n4.nabble.com/SSL-handshake-stalls-broker-with-NIO-td4677916.html
> - http://stackoverflow.com/questions/19395769/activemq-mqtt-maxes-cpu-on-first-client-connect
>  


--
This message was sent by Atlassian JIRA
(v6.3.4#6332)