You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@activemq.apache.org by mikebroberts <mi...@gmail.com> on 2007/10/01 19:33:00 UTC
Re: Out of memory error with durable subscribers
Is there any chance someone from the ActiveMQ team can give some insiders
feedback on the below? Even if it's 'durable subscribers are unstable -
don't use them'?
Thanks,
Mike Roberts (development manager for the team Ahmed is on)
aselshamy wrote:
>
> Hi
>
> We have one topic publisher that publish 2M messages with 4 durable
> subscribers. Things go fine till we reached 1.7M we get the
> outofmemoryerror.
> AMQ is setup to use 5 * 20M log files, and derbydb jdbc persistence. It
> starts with -Xmx1024M and <usageManager id="memory-manager" limit="512
> MB"/>
>
> At the time of the out of memory, one of the log file is gone crazy and
> it's size is 415M, derbydb size is so big as well.
>
> We are using session client acknowledgment. and prefetch size of 1 (we
> need to serialize message consumption).
>
> each message is being acknowledged using javax.jms.Message.acknowledge();
>
> did not find anything on how to change checkpoint interval for
> persistence.
>
> Any thoughts on what could be the issue?
>
--
View this message in context: http://www.nabble.com/Out-of-memory-error-with-durable-subscribers-tf4487985s2354.html#a12983727
Sent from the ActiveMQ - User mailing list archive at Nabble.com.
Re: Out of memory error with durable subscribers
Posted by aselshamy <as...@hotmail.com>.
We did not do a jvm kill -3, we got the OOME on the broker. We were able to
send more messages because we pushed the memory size to 1024M for jvm, 512M
for broker to allow more time for the check point to occur (we were guessing
that was the issue, but it was not).
BTW we are using AMQ version 4.1.1.
--
View this message in context: http://www.nabble.com/Out-of-memory-error-with-durable-subscribers-tf4487985s2354.html#a12986052
Sent from the ActiveMQ - User mailing list archive at Nabble.com.
Re: Out of memory error with durable subscribers
Posted by WHIRLYCOTT <ph...@whirlycott.com>.
Mike,
Are you getting OOMEs on the publisher (the client) or in the broker?
I'm wondering if you can send your jvm a kill -3 signal to get a
thread dump, because it sounds vaguely like the problem that I'm
currently experiencing. However, my publishers can only send ~1390
messages before getting an OOME with a 64Mb heap.
phil.
On Oct 1, 2007, at 1:33 PM, mikebroberts wrote:
>
> Is there any chance someone from the ActiveMQ team can give some
> insiders
> feedback on the below? Even if it's 'durable subscribers are
> unstable -
> don't use them'?
>
> Thanks,
>
> Mike Roberts (development manager for the team Ahmed is on)
>
>
> aselshamy wrote:
>>
>> Hi
>>
>> We have one topic publisher that publish 2M messages with 4 durable
>> subscribers. Things go fine till we reached 1.7M we get the
>> outofmemoryerror.
>> AMQ is setup to use 5 * 20M log files, and derbydb jdbc
>> persistence. It
>> starts with -Xmx1024M and <usageManager id="memory-manager"
>> limit="512
>> MB"/>
>>
>> At the time of the out of memory, one of the log file is gone
>> crazy and
>> it's size is 415M, derbydb size is so big as well.
>>
>> We are using session client acknowledgment. and prefetch size of
>> 1 (we
>> need to serialize message consumption).
>>
>> each message is being acknowledged using
>> javax.jms.Message.acknowledge();
>>
>> did not find anything on how to change checkpoint interval for
>> persistence.
>>
>> Any thoughts on what could be the issue?
>>
>
> --
> View this message in context: http://www.nabble.com/Out-of-memory-
> error-with-durable-subscribers-tf4487985s2354.html#a12983727
> Sent from the ActiveMQ - User mailing list archive at Nabble.com.
>
--
Whirlycott
Philip Jacob
phil@whirlycott.com
http://www.whirlycott.com/phil/