Re: Weird Message Driven Bean Behavior
Nigel Deakin wrote:
>> On the other side we have a JEE app (in the same glassfish server)
>> which has
>> a MDB to get the message and process it... All was fine but then a
>> week ago
>> we started seeing some weird issue where the MessageProducer is
>> putting a
>> message in the queue and it takes up to 4 seconds for the MDB to grab
> Remember you've got a pool of MDB instances running in your GlassFish
> server. If they're all busy processing messages then when a new
> message arrives it can't be processed until a MDB instance becomes
GlassFish will create as many instances as necessary, but there can be
not enough threads, or the JCA adapter might be capping the amount of
messages it processes at once.
> Do you see this problem when the load is very low and there are lots
> of free MDB instances?
> When you see the problem, does it continue for ever or does it go away
> after a while?
>> One other thing, the MDB is not in a regular EJB jar, it is in a
>> hybrid osgi
> That's probably not the cause of the issue.
>> We are trying to recreate the problem in our performance test
>> envrionment but
>> we cannot recreate it. Has anyone seen this or have any clue if we
>> need to
>> fix some parameter to prevent this from happening?
> A standard piece of advice is to check both the GlassFish server log
> and the JMS broker log for unusual entries (Say if you don't know how
> to find the latter).