You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@activemq.apache.org by "Rodrigo Sampaio Vaz (JIRA)" <ji...@apache.org> on 2010/07/01 00:14:51 UTC

[jira] Created: (AMQ-2808) connection blocked when using stomp or stomp+ssl

connection blocked when using stomp or stomp+ssl
------------------------------------------------

                 Key: AMQ-2808
                 URL: https://issues.apache.org/activemq/browse/AMQ-2808
             Project: ActiveMQ
          Issue Type: Bug
          Components: Broker, Transport
    Affects Versions: 5.3.2
         Environment: Linux 64 bit, debian, sun java 1.6.0_20
            Reporter: Rodrigo Sampaio Vaz


When a client opens a connection and unexpectedly finishes the connection remains open and established on the activemq machine, follow are the steps to reproduce:

1 - Start the activemq using the stomp connector or stomp+ssl 
2 - Use tcpkill -9 port 61612 (default por for stomp connector)
3 - From another machine send a telnet connection to the stomp port like telnet <stomp_server_ip> <stomp_port>, this connection is going to be finished by tcpkill
4 - On the activemq machine issue a netstat -pont | grep <stomp_port>
You should  see that a connection is still ESTABLISHED and the number of connections grows if you start more telnets to the port. To the activemq server it thinks that the clients are still connected even if they are dead.
This does not happens when using stomp+nio, but since there is no ssl over stomp+nio we loose the encryption.


-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


[jira] Issue Comment Edited: (AMQ-2808) connection blocked when using stomp or stomp+ssl

Posted by "James Green (JIRA)" <ji...@apache.org>.
    [ https://issues.apache.org/activemq/browse/AMQ-2808?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=61041#action_61041 ] 

James Green edited comment on AMQ-2808 at 8/3/10 10:12 AM:
-----------------------------------------------------------

I think I'm seeing this issue.

For me I have a test script I can run at the bash shell or via apache. If I repeatedly run and then ctrl+c the script from bash it reconnects no problem. If I run via Apache, then hit the browser stop button, then reload I instantly get told that my client-id is already connected.

Netstat -nat hows the client connection remains ESTABLISHED.

I am using Ubuntu 10.04 32 bit with ActiveMQ 5.3.2.

Just to add that I am connecting using stomp over ssl.

      was (Author: jmkgreen):
    I think I'm seeing this issue.

For me I have a test script I can run at the bash shell or via apache. If I repeatedly run and then ctrl+c the script from bash it reconnects no problem. If I run via Apache, then hit the browser stop button, then reload I instantly get told that my client-id is already connected.

Netstat -nat hows the client connection remains ESTABLISHED.

I am using Ubuntu 10.04 32 bit with ActiveMQ 5.3.2.

  
> connection blocked when using stomp or stomp+ssl
> ------------------------------------------------
>
>                 Key: AMQ-2808
>                 URL: https://issues.apache.org/activemq/browse/AMQ-2808
>             Project: ActiveMQ
>          Issue Type: Bug
>          Components: Broker, Transport
>    Affects Versions: 5.3.2
>         Environment: Linux 64 bit, debian, sun java 1.6.0_20
>            Reporter: Rodrigo Sampaio Vaz
>
> When a client opens a connection and unexpectedly finishes the connection remains open and established on the activemq machine, follow are the steps to reproduce:
> 1 - Start the activemq using the stomp connector or stomp+ssl 
> 2 - Use tcpkill -9 port 61612 (default por for stomp connector)
> 3 - From another machine send a telnet connection to the stomp port like telnet <stomp_server_ip> <stomp_port>, this connection is going to be finished by tcpkill
> 4 - On the activemq machine issue a netstat -pont | grep <stomp_port>
> You should  see that a connection is still ESTABLISHED and the number of connections grows if you start more telnets to the port. To the activemq server it thinks that the clients are still connected even if they are dead.
> This does not happens when using stomp+nio, but since there is no ssl over stomp+nio we loose the encryption.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


[jira] Commented: (AMQ-2808) connection blocked when using stomp or stomp+ssl

Posted by "James Green (JIRA)" <ji...@apache.org>.
    [ https://issues.apache.org/activemq/browse/AMQ-2808?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=61041#action_61041 ] 

James Green commented on AMQ-2808:
----------------------------------

I think I'm seeing this issue.

For me I have a test script I can run at the bash shell or via apache. If I repeatedly run and then ctrl+c the script from bash it reconnects no problem. If I run via Apache, then hit the browser stop button, then reload I instantly get told that my client-id is already connected.

Netstat -nat hows the client connection remains ESTABLISHED.

I am using Ubuntu 10.04 32 bit with ActiveMQ 5.3.2.


> connection blocked when using stomp or stomp+ssl
> ------------------------------------------------
>
>                 Key: AMQ-2808
>                 URL: https://issues.apache.org/activemq/browse/AMQ-2808
>             Project: ActiveMQ
>          Issue Type: Bug
>          Components: Broker, Transport
>    Affects Versions: 5.3.2
>         Environment: Linux 64 bit, debian, sun java 1.6.0_20
>            Reporter: Rodrigo Sampaio Vaz
>
> When a client opens a connection and unexpectedly finishes the connection remains open and established on the activemq machine, follow are the steps to reproduce:
> 1 - Start the activemq using the stomp connector or stomp+ssl 
> 2 - Use tcpkill -9 port 61612 (default por for stomp connector)
> 3 - From another machine send a telnet connection to the stomp port like telnet <stomp_server_ip> <stomp_port>, this connection is going to be finished by tcpkill
> 4 - On the activemq machine issue a netstat -pont | grep <stomp_port>
> You should  see that a connection is still ESTABLISHED and the number of connections grows if you start more telnets to the port. To the activemq server it thinks that the clients are still connected even if they are dead.
> This does not happens when using stomp+nio, but since there is no ssl over stomp+nio we loose the encryption.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.