activemq-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Christian Posta <christian.po...@gmail.com>
Subject Re: Activemq producer locks
Date Mon, 15 Apr 2013 14:44:28 GMT
Interesting. Sounds like a TCP simultaneous connect maybe?

http://stackoverflow.com/questions/2231283/tcp-two-sides-trying-to-connect-simultaneously




On Sun, Apr 14, 2013 at 10:40 PM, arjun <arjun.418@gmail.com> wrote:

> Hi,
>
> A few more recent observations.
>
> consider this -
>
> we have a message producer that uses a failover transport
> failover://(tcp://host1:61616,tcp://host2:61616).
>
> there are 2 brokers - 1 on host 1 and other on host2 and both listen to
> 61616 on the respective machines. Both point to same data store(oracle db)
> and while executing the test case, the broker on host 2 is the one that has
> acquired the db lock and hence is the active instance
>
> Now, when the message producer gets 61616 as the ephemeral port to make the
> tcp connection, the thread seems to hang. an lsof listing of activity on
> port 61616(lsof -iTCP:61616) displays this -
>
> java the_pid_of_the_message_producer .. all other info and then this is the
> name -
>
>  host1:61616->host1:61616
>
> The port is never released by the thread. Also, in this state, when the
> broker on host1 tries to become the active broker(if we bring down the
> broker on host2), it fails to start up with the address already in use.
>
> Could someone please explain whats happening here. Is the solution for this
> to make sure 61616 is not got as an ephemeral port on host1(I am not able
> to
> figure why that would be necessary as there is nothing listening on port
> 61616 on host1. We are using AMQ 5.5.0
>
>
>
> --
> View this message in context:
> http://activemq.2283324.n4.nabble.com/Activemq-producer-locks-tp4660619p4665920.html
> Sent from the ActiveMQ - User mailing list archive at Nabble.com.
>



-- 
*Christian Posta*
http://www.christianposta.com/blog
twitter: @christianposta

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message