Follow up:
removing the parameter queuePrefetch=10000 from the RA solves that.
Even if I don't fully understand the reason, it seems not to complain
anymore.
drvillo wrote:
>
> Hi James
> I've just ended some testing with mysql.
> Throughput has moved from 30 msg/sec to almost 300 (niice),
> but I've started to see the following messages:
>
> [.0.1:58512] INFO org.apache.activemq.broker.region.PrefetchSubscription
> - Could not correlate acknowledgment with dispatched message: MessageAck
> {commandId = 300079, responseRequired = false, ackType = 2, consumerId =
> ID:villomac.local-58504-1168608802249-3:1:-1:2, firstMessageId =
> ID:villomac.local-58504-1168608802249-3:4:10388:1:1, lastMessageId =
> ID:villomac.local-58504-1168608802249-3:4:10388:1:1, destination =
> queue://queue2, transactionId =
> TX:ID:villomac.local-58504-1168608802249-3:1:99999, messageCount = 1}
>
>
> The setup is the same, with the only difference of the datasource...
> does it make any sense? Or should I just ignore them?
>
> FYI, I'm using DUPS_OK_ACKNOWLEDGE.
>
> Anyway thanks, now I'll test it with hsqldb.
> Cheers
> Francesco
>
> James.Strachan wrote:
>>
>> On 1/12/07, drvillo <[EMAIL PROTECTED]> wrote:
>>>
>>> HI James
>>>
>>> no I haven't tried, since my specs require to have all embedded, but if
>>> it's
>>> for isolating the problem I can try.
>>> Do you think hsqldb would do better?
>>
>> I don't think so but by all means try it if you like. I've never
>> really considered hsqldb as a 'real' database for massive amounts of
>> data ;) I figured folks just used it for testing.
>>
>> If nothing else, trying mysql / postgresql or some other external
>> database (preferably on the same box) and it would at least narrow
>> down the problem. I've heard reports of Derby hogging RAM and slowing
>> down with large databases.
>>
>> --
>>
>> James
>> -------
>> http://radio.weblogs.com/0112098/
>>
>>
>
>
--
View this message in context:
http://www.nabble.com/performance-degradation-tf2925875.html#a8298816
Sent from the ActiveMQ - User mailing list archive at Nabble.com.
