[beve]

Hi Anders,

I've been able to reproduce this using the main trunk. When this happens I get alot of messages in both the console and server.log.
I noticed in server.log :

2008-12-30 13:32:36,205 DEBUG [org.jboss.internal.soa.esb.rosetta.pooling.JmsConnectionPool] The connection pool was exhausted, waiting for a session to be released.

Can you check your server.log and see if you are getting the above debug statement about the jms connection pools being exhausted?

I've created a jira for this so we can track it:
http://www.jboss.com/index.html?module=bb&op=viewtopic&t=147723

Thanks,

/Daniel

[anders.andersen]

Hi Daniel,

Yeah, that is correct - I just looked in the log from a new test run I did and the log has very many:

2008-12-30 14:04:08,858 DEBUG [org.jboss.internal.soa.esb.rosetta.pooling.JmsConnectionPool] The connection pool was exhausted, waiting for a session to be released.

Think we might be zeroing in on the core issue here! Somehow the connection pool gets empty very fast and we are forced to wait for a new session.

Thanks for the quick reply! ... will be looking forward to see the outcome of this ;-)


Kind regards,

Anders Andersen.

[beve]

Hi Anders,

can you try increasing the following value in jbossesb.sar/jbossesb-properties.xml:

<property name="org.jboss.soa.esb.jms.connectionPool" value="20"/>

I was able to re-run the test without failure when settning it to 100.

regards,

/Daniel

[anders.andersen]

Hi Daniel,

I will try this first thing Friday morning the 2/1 (gone home for today). Sounds like a solution, but we are not just "pushing the problem in front of us" by raising the number of connections in the pool? It might be a connection-leak... Anyways, I will try this as soon as possible...


Kind regards and a happy new year,

Anders Andersen.

[anders.andersen]

Hi Daniel,

I tried your suggestion and it works perfect here! I have even tried running with 150 threads and 10,000 invocations and it still runs nicely even though the average response time raises to nearly 4 sec (incl. a lot of logging and monitoring). But this is perfectly OK, since we are putting a lot of load on the server (with just 1 thread the average is about 35 ms for me, and with 15 threads I get about 800 ms). Now, I have not found any problems over time (running a soak-like test), thus we might conclude that this solution effectively solves the problem?! And if so, will it be included in the next JBossESB release (when it is ready)?

Regarding the other issue, "Badly formed EPR", this is not a big problem for us, since it does not really have any performance and/or business impact, but it should maybe be addressed in the future.

Once again, thanks a lot for your quick and resourceful help on this issue!


Kind regards,

Anders Andersen.

[beve]

Hi Anders,

And if so, will it be included in the next JBossESB release (when it is ready)?

The property 'org.jboss.soa.esb.jms.connectionPool' is actually available in the latest release 4.4.GA. But if you are referring to a fix for the 'Badly formed EPR' hopefully it will be in for the next release.

Regarding the other issue, "Badly formed EPR", this is not a big problem for us, since it does not really have any performance and/or business impact, but it should maybe be addressed in the future.

Yeah, I agree. I'll add a jira for this and make sure it gets seen to.

nps, and thanks for reporting this!

/Daniel

[beve]

Here is the Jira for tracking the "Badly formed EPR" issue:
https://jira.jboss.org/jira/browse/JBESB-2258

[kconner]

The unresponsive EPR issue is likely to be https://jira.jboss.org/jira/browse/JBESB-1929

The service invoker was, incorrectly, holding on to two couriers when it should have been releasing the first prior to obtaining the second.

It was, therefore, possible for all threads to have acquired the first without being able to obtain the second.

A workaround is to increase the pool size, as you have already noted.

This fix is already included in trunk and will be release as part of ESB 4.5.

Kev

[anders.andersen]

Hi,

Great news that the new 4.5 is out! However I'm a little confused if this issue has been fixed or not in the new release.https://jira.jboss.org/jira/browse/JBESB-2255 says the issue is:

Status: Resolved
Resolution: Rejected

Does this mean that the issue is fixed or not?


Kind regards

Anders Andersen

[beve]

Hi Anders,

Does this mean that the issue is fixed or not?

In this case of https://jira.jboss.org/jira/browse/JBESB-1929 this was rejected as there was nothing to fix. We added a "Workaround Description" to the case which is the "fix". This is something that would be configured to meet individual deployment needs.

The jira that Kev mentioned (https://jira.jboss.org/jira/browse/JBESB-1929) has been fixed an included in 4.5.
So you might try lowering the value of org.jboss.soa.esb.jms.connectionPool that you have now and run your test with 4.5 and see if you where running into JBESB-1929.

regards,

/Daniel