users@glassfish.java.net

IIOP leak was: Re: OpenFileDescriptor leak in GFv2UR1?

From: Jeanfrancois Arcand <Jeanfrancois.Arcand_at_Sun.COM>
Date: Thu, 06 Mar 2008 11:30:54 -0500

Hi,

can you file an issue here under category 'orb'

https://glassfish.dev.java.net/servlets/ProjectIssues

from the netstat -an, looks like port 3700 (IIOP) is leaking.

Thanks!

-- Jeanfrancois

Witold Szczerba wrote:
> 2008/3/6, Jeanfrancois Arcand <Jeanfrancois.Arcand_at_sun.com>:
>> Hi Witold
>>
>> Witold Szczerba wrote:
>> > Hi there,
>> > have a look at the picture I've attached. It is showing GFv2UR1 in our
>> > customer's production environment running our application (EJB3
>> > local+remote and Swing client).
>> >
>>
>> No http in the path? Just RMI/IIOP for reaching EJB, right? That will
>> clear the http path :-)
>
> There is actually one pure EJB(RMI/IIOP) EAR and one WAR with nothing
> more than JNLPDownload servlet, but I'm 99,9% sure it has nothing to
> do with that servlet, as usually people are launching application in
> the morning (and launching application is the only interaction with
> that servlet), where as the chart shows leaking during the entire day.
>
>> > The chart starts on Thursday morning and goes up, then just after 5pm,
>> > when remote clients disconnects (people go home) it goes down a little
>> > bit, during night it is stable, then it grows again during Friday up
>> > until somewhere 5pm. Then weekend begins and line is almost completely
>> > horizontal until Monday morning... as you can see during this week
>> > everything looks pretty much the same.
>> >
>> > I am not sure, but it looks like some kind of leak.
>>
>> Agree.
>>
>>
>> I was reporting
>> > similar behavior about year ago against GFv2 beta
>>
>> Have you filed an issue? Which #
>
> Oh my, now I see I did not fill an issue, but here is our previous
> conversation about this:
> http://forums.java.net/jive/thread.jspa?messageID=224477
> I was waiting until final GFv2 release and when it came, we were
> unable to use it because of another, unrelated issue (#3683 - thank
> you for resolving) and we were using that old beta version up until a
> week ago.
>
>
>> > The problem is I cannot reproduce this issue in testing environment.
>> > There are about 150 users using our application deployed on that
>> > production server.
>> > Maybe someone from Glassfish team have some idea what it could be?
>>
>>
>> It is hard to debug here as we cannot see which part is leaking. I would
>> try to do a netstat -an and post the result here...should be clear with
>> that information where the leak is coming from.
>
> netstat -an is attached.
> About enviroment:
> Red Hat 5, uname -a:
> Linux mikro 2.6.18-53.el5 #1 SMP Wed Oct 10 16:34:19 EDT 2007 x86_64
> x86_64 x86_64 GNU/Linux
> java version "1.6.0_04"
> Java(TM) SE Runtime Environment (build 1.6.0_04-b12)
> Java HotSpot(TM) Server VM (build 10.0-b19, mixed mode)
>
>> Thanks
>> -- Jeanfrancois
>
> My pleasure, thanks!
> Witold Szczerba
>
>
> ------------------------------------------------------------------------
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe_at_glassfish.dev.java.net
> For additional commands, e-mail: users-help_at_glassfish.dev.java.net