users@glassfish.java.net

Re: HTTP listener threads are not returned to thread pool after an error

From: Oleksiy Stashok <Oleksiy.Stashok_at_Sun.COM>
Date: Mon, 19 Apr 2010 11:43:57 +0200

Hi,

can you pls. take a threads snapshot [1] and attach it here?

Thanks.

[1] http://blogs.sun.com/oleksiys/entry/glassfish_hangs
On Apr 19, 2010, at 11:26 , glassfish_at_javadesktop.org wrote:

> Hi,
>
> I see some weird behavior in our GF 2.1 env. I have an EAR app
> deployed on it, the app contains number of stateless session beans,
> accessed externally as webs services. In case when any error happens
> during WS request processing HTTP thread assigned to the request is
> not returned to the thread pool. In the server log I see that server
> tries to perform request again and again (I don't have any retry
> behavior in my code). When number of busy threads exhausted thread
> pool size the server stops to respond to any incoming call. After
> restart it works for some time, but then the same scenario happens
> again. The server log looks like:
>
> [#|2010-04-17T13:40:02.737+0400|WARNING|sun-appserver2.1|
> my.services.impl.DeviceProfileClientServiceImpl|
> _ThreadID
> =
> 221
> ;_ThreadName
> =
> httpSSLWorkerThread
> -8080-127;_RequestID=e7d9f6ca-9d5c-41b9-87f8-6e3244436106;|Device
> Profile Service threw an exception: [null]. Ignoring...|#]
>
> [#|2010-04-17T13:41:12.047+0400|SEVERE|sun-appserver2.1|
> javax.enterprise.resource.corba|
> _ThreadID
> =
> 221
> ;_ThreadName
> =
> httpSSLWorkerThread
> -8080-127;_RequestID=e7d9f6ca-9d5c-41b9-87f8-6e3244436106;|IOP5041:
> Principal propagation: Cannot find principal information in subject|#]
> ..................
> [#|2010-04-17T16:14:42.087+0400|SEVERE|sun-appserver2.1|
> javax.enterprise.resource.corba|
> _ThreadID
> =
> 221
> ;_ThreadName
> =
> httpSSLWorkerThread
> -8080-127;_RequestID=e7d9f6ca-9d5c-41b9-87f8-6e3244436106;|IOP5041:
> Principal propagation: Cannot find principal information in subject|#]
>
> See the same ThreadID and RequestID in the log records. (Don't pay
> attention on the error msgs, they're expected). The server was not
> restarted during this time, so the thread was busy for > 2 hours...
>
> I don't have such issue when service invoked via remote EJB call.
>
> Any idea on how to avoid this problem?
>
> Thanks, Denis.
> [Message sent by forum member 'sdv0967']
>
> http://forums.java.net/jive/thread.jspa?messageID=397759
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe_at_glassfish.dev.java.net
> For additional commands, e-mail: users-help_at_glassfish.dev.java.net
>