There is nothing out of the ordinary in server.log that I am aware of.
The NP Pool column in Task Manager for java.exe is fairly stable most of
the time, but once whatever happens, it takes minutes or possibly a
couple of hours for it to take down the system. Since December I
haven't been able to do diagnostics while it is in this state because
the OS is completely crashed. We have to cut the power to reboot it.
Ryan
Martin Gainty wrote:
> Ryan-
>
> any information in either \domains\domain1\logs\server.log
> \domains\domain1\logs\jvm.log?
>
> Anything from jconsole..(check for spike in heap)
>
> ?
> Martin
> ______________________________________________
> Disclaimer and confidentiality note
> Everything in this e-mail and any attachments relates to the official
> business of Sender. This transmission is of a confidential nature and
> Sender does not endorse distribution to any party other than intended
> recipient. Sender does not necessarily endorse content contained
> within this transmission.
>
>
>
>
> > Date: Mon, 26 Jan 2009 16:12:28 -0500
> > From: ryan_at_ijws.com
> > To: users_at_glassfish.dev.java.net
> > Subject: Re: GlassFish on Windows Server 2008
> >
> > Regarding test code to verify this behavior, there might be. It is very
> > difficult to reproduce (except on our server) and that is why after
> > almost 2 years Microsoft has not fixed it. Last year someone at Sun
> > told me that another customer provided a full application that runs on
> > GlassFish and can help them reproduce this issue. I would think Sun
> > could take that same app and try it on Windows 2008 server to check if
> > the problem is still there.
> >
> > We still have absolutely no idea what triggers the problem on our
> > server. It seems to be fine for days and days, thousands of
> > transactions. Out of the blue Windows just crashes (blank grey screen
> > at console, can't connect to it remotely, but you can ping it). A
> > couple of times I was in just before that happened and saw NP Pool in
> > task manager climbing continually, above 10000K. I still think it is
> > somehow related to communication with SQL server over the network,
> > because a different JDBC driver used to record "TDS prelogin response
> > error" before opening the NP Pool flood gates.
> >
> >
> > Ryan
> >
> >
> >
> > Ryan de Laplante wrote:
> > > It was me who contacted Microsoft about that case and tried the
> > > HotFix. There were three (third was a change to the registry) and I
> > > can confirm that it did not fix the problem. It seemed to get worse,
> > > but I can't confirm it was because of the hotfix or increased
> > > usage. We used to be able to deal with this problem by rebooting
> > > once a month or so, but since December 2008 it is so bad that Windows
> > > completely crashes (requires hard reset) about twice per week. We're
> > > in the process of planning a switch to Red Hat Enterprise Linux, but
> > > would prefer Windows 2008 Server if it is 100% guaranteed not to
> > > happen on that OS. Jeanfrancois @ Sun says the problem happens on
> > > Windows 2000 and Windows 2003. Not just 2003.
> > >
> > >
> > > Thanks,
> > > Ryan
> > >
> > >
> > > NBW wrote:
> > >> I read in an old thread on this subject that the HotFix did not fix
> > >> the problem and some folks reported it got worse. Would like to
> > >> verify that this is not an issue under Windows Server 2008 as well.
> > >> Is there any test code that exercises this issue available?
> > >>
> > >> Thanks.
> > >>
> > >> On Mon, Jan 26, 2009 at 3:13 PM, Paul Sterk <Paul.Sterk_at_sun.com
> > >> <mailto:Paul.Sterk_at_sun.com>> wrote:
> > >>
> > >>
> > >> Ryan,
> > >>
> > >> I checked the status of bug 6575349 and there is a note that the
> > >> bug was only found on Windows 2003. There is an entry in the 9.1pe
> > >> release notes about downloading a HotFix from Microsoft:
> > >>
> > >> Case Number: SRX070906600011
> > >> Workspace URL:
> > >>
> > >>
> (https://sftus.one.microsoft.com/ChooseTransfer.aspx?key=a9c9b140-aa2f-4e0d-a39c-7e89ba067af6)
>
> > >>
> > >>
> > >> HTH,
> > >> Paul
> > >> GlassFish Community Manager
> > >>
> > >>
> > >> On 1/23/2009 11:06 PM, Ryan de Laplante wrote:
> > >>
> > >> Hi,
> > >>
> > >> We are one of the unlucky customers suffering from this
> > >> problem on
> > >> Windows 2003 Server:
> > >>
> > >>
> > >> Win2003 only: Non-paged pool leak memory, breaking tcp
> > >> stack and
> > >> richaccess test (6575349)
> > >>
> > >>
> > >> Description
> > >>
> > >> *(Windows 2003 only)* There are memory leaks on Windows
> > >> 2003 systems
> > >> when performing rich access functions. The problem occurs
> > >> because the
> > >> Win32 nonpaged pool keeps growing, eventually bringing
> > >> down the entire
> > >> TCP/IP stack. Once the failure happens, the TCP/IP stack
> > >> is left in an
> > >> recoverable state, and the only way restore it is by
> > >> rebooting the
> > >> Windows 2003 system.
> > >>
> > >>
> > >> Workaround
> > >>
> > >> There are two workarounds to this issue:
> > >>
> > >> *
> > >>
> > >> Use Grizzly blocking mode by configuring the domain.xml
> > >> http-listener attribute, blocking-enabled="true" or add the
> > >> following http-listener property:
> > >>
> > >>
> > >> <property name="blocking" value="true"/>
> > >>
> > >> *
> > >>
> > >> Use Windows Vista or Windows XP.
> > >>
> > >>
> > >> I noticed that GlassFish V2.1 is certified for use on Windows
> > >> 2008
> > >> server. Does that mean Sun and Microsoft have confirmed that
> > >> the problem
> > >> does not exist on Windows 2008? If so, they must have found
> > >> the source
> > >> of the problem and created a patch for Windows 2003?
> > >>
> > >>
> > >> Thanks,
> > >> Ryan