You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@activemq.apache.org by Tito Flagella <ti...@link.it> on 2006/07/11 23:52:38 UTC
durable topic
We are testing AMQ 4.0.1 in the following scenario:
- a durable topic;
- a producer continuosly pushing some thousands of messages of various
sizes (1, 10 and 100k);
- no active consumer.
We are doing that in order to test the case in which the consumers
could be offline for some time. In this condition, we expect AMQ be
able to go on regularly accepting new produced messages, only limited by
the available disk space.
Unfortunatly we are experiencing an inefficient broker behavior during
the test: the broker used heap memory grows up to the maximum allowed,
then the broker stops accepting other messages, closes all open
connections and starts a very long recovery operation (10 minutes on our
test server), during which the used heap memory decreases to the initial
one. Then the broker starts working again, and the memory to grow, until
the next recovery operation.
Is there any way we can configure the topic parameters or the broker
itself, to avoid this long periods of broker unavailability when
consumers are offline?
Thanks a lot,
tito.
Re: durable topic
Posted by Tito Flagella <ti...@link.it>.
On Wed, 12 Jul 2006, James Strachan wrote:
> Do you have a test case that can rreproduce this?
The problem disappeared when using postgres as datasource for durable
topics.
tito.
Re: durable topic
Posted by James Strachan <ja...@gmail.com>.
Do you have a test case that can rreproduce this?
On 7/11/06, Tito Flagella <ti...@link.it> wrote:
> We are testing AMQ 4.0.1 in the following scenario:
>
> - a durable topic;
> - a producer continuosly pushing some thousands of messages of various
> sizes (1, 10 and 100k);
> - no active consumer.
>
> We are doing that in order to test the case in which the consumers
> could be offline for some time. In this condition, we expect AMQ be
> able to go on regularly accepting new produced messages, only limited by
> the available disk space.
>
> Unfortunatly we are experiencing an inefficient broker behavior during
> the test: the broker used heap memory grows up to the maximum allowed,
> then the broker stops accepting other messages, closes all open
> connections and starts a very long recovery operation (10 minutes on our
> test server), during which the used heap memory decreases to the initial
> one. Then the broker starts working again, and the memory to grow, until
> the next recovery operation.
>
> Is there any way we can configure the topic parameters or the broker
> itself, to avoid this long periods of broker unavailability when
> consumers are offline?
>
> Thanks a lot,
>
> tito.
>
>
>
--
James
-------
http://radio.weblogs.com/0112098/