On 1/12/12 2:46 PM, forums_at_java.net wrote:
> We are running Glassfish 3.1.1 build 12 on Windows Server 2008 R2
> Enterprise,
> 4 GB RAM, 64-bit operating system, 4 processors, 2.93GHz. This is our
> first
> experiencing running an application server. We have an application
> deployed to Glassfish that is basically a handler of printing task (i.e.
> application name is Taskserver). Basically consist of JSP, Servlets,
> EJB (session and mdb), utilizes JMS and has the basic configured
> administered resources in Glassfish (JDBC, JMS, connection pools, etc.).
>
> The application basically does the following:
>
> 1) Reads entries from an AS400 data queue at a regularly interval
> (polling
> frequency is 2000 milliseconds)
> 2) These entries are submitted to the data queue by our endusers when
> they
> request the printing of a shipping document (i.e. picklist,
> packlist, BOL, invoices, etc.).
> 3) The data queue entries are used to create messages that are
> consumed by
> an MDB. The content of the message is the information related to the
> requested print task.
> 4) The message is parsed by the MDB and the task_ID is used to determine
> which bean's print method to call (i.e. PickListPrintMgr,
> PackListPrintMgr, etc.)
> 5) These print manager beans get a connection to the AS400 which is
> used (1)
> when retrieving data from the ERP database related to the sales order
> and (2) when spooling the output back to the AS400 output queue.
>
> Here is what is perplexing. When volume testing this application, I will
> submit over 1000 request to print a pick list. This will either come in
> one request or multiple request submitted almost
> simulatenously. Glassfish
> will hang and processing will stop when around 1000 of these request have
> been processed (have generated a file in my output queue). I've been
> researching this issue and am curious if anyone has experienced a similar
> problem? I have turned on monitoring within Glassfish and have been
> examing
> those logs, also looking for issues in the server log, but I've not
> really
> sure what I should be looking for. Anyone have any input that might be
> helpful? Where to look, what to bee looking for? I have bumped up the
> max number of connections for the connection pool, EJB and MDB settings,
> but this has not seemed to have any real effect.
It would probably be a good idea to get a thread dump of your server
when it's in
it's hung state.
Byron has a good blog [1] on doing this from the command line.
You could also use a tool like Visual VM to obtain a thread dump as well.
Once you've got the thread dump, please post it here and we can try to help.
[1]
http://blogs.oracle.com/foo/entry/how_to_get_thread_dumps
>
> Thanks in advance,
> Randy
>
>
> --
>
> [Message sent by forum member 'Randy_Dennis']
>
> View Post: http://forums.java.net/node/882682
>
>