You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@activemq.apache.org by beyond005 <be...@gmail.com> on 2008/08/04 03:41:33 UTC
How to deal with blocked transport
There are some comments about blocked transport in slow consumer topic:
when using TCP there can be occasions when a network outage can result in a
blocked write. This can cause the entire broker to freeze - and the socket
may never be unblocked. Currently we have a Thread that checks for blocked
sockets - using a sweep protocol to detect if there are connections that are
blocked writing for more than a configurable period. This can work but there
is no way to unblock the calling thread that is associated with the send
down the socket (which will be the publishing thread in non-durable topics).
Could anybody give a feasible way to resolve this problem in a product
environment. Currently we just can reboot the whole activemq to restart the
socket.
BTW, we are using 4.1.1 version of activemq in our environment. Could I
assume that activemq 5.1 has beat it?
--
View this message in context: http://www.nabble.com/How-to-deal-with-blocked-transport-tp18804433p18804433.html
Sent from the ActiveMQ - User mailing list archive at Nabble.com.
Re: How to deal with blocked transport
Posted by Hiram Chirino <hi...@hiramchirino.com>.
5.1 Does have better dead connection handling. Please test it out and
let us know if you see the improvements.
On Sun, Aug 3, 2008 at 9:41 PM, beyond005 <be...@gmail.com> wrote:
>
> There are some comments about blocked transport in slow consumer topic:
>
> when using TCP there can be occasions when a network outage can result in a
> blocked write. This can cause the entire broker to freeze - and the socket
> may never be unblocked. Currently we have a Thread that checks for blocked
> sockets - using a sweep protocol to detect if there are connections that are
> blocked writing for more than a configurable period. This can work but there
> is no way to unblock the calling thread that is associated with the send
> down the socket (which will be the publishing thread in non-durable topics).
>
> Could anybody give a feasible way to resolve this problem in a product
> environment. Currently we just can reboot the whole activemq to restart the
> socket.
>
> BTW, we are using 4.1.1 version of activemq in our environment. Could I
> assume that activemq 5.1 has beat it?
> --
> View this message in context: http://www.nabble.com/How-to-deal-with-blocked-transport-tp18804433p18804433.html
> Sent from the ActiveMQ - User mailing list archive at Nabble.com.
>
>
--
Regards,
Hiram
Blog: http://hiramchirino.com
Open Source SOA
http://open.iona.com