[THIN] Re: What happens when you overload a Citrix server?

  • From: "Landin, Mark" <Mark.Landin@xxxxxxxxxxxxxxxx>
  • To: "thin@xxxxxxxxxxxxx" <thin@xxxxxxxxxxxxx>
  • Date: Thu, 4 Nov 2010 09:13:43 -0500

Warren,

Thanks for this info. Can you explain a little bit more what this hook is, and 
what effect there is in turning it off? I'd like to understand the impact of 
this change a little more before implementing it.

I should point out that we have been running GP in our Citrix environment since 
2002 (started with GP 6 on PS 3, then GP 7.5 on PS 4, now GP 10 on PS 4.5) and 
we've only been seeing this crashing behavior for the last 3 or 4 months, after 
we added the manufacturing module to GP and added about 2 dozen new users 
because of it. This is what makes us think it's probably a server resource 
issue first.

-----Original Message-----
From: thin-bounce@xxxxxxxxxxxxx [mailto:thin-bounce@xxxxxxxxxxxxx] On Behalf Of 
Warren Simondson
Sent: Wednesday, November 03, 2010 4:51 PM
To: thin@xxxxxxxxxxxxx
Subject: [THIN] Re: What happens when you overload a Citrix server?

It could be the MS dynamics program that is causing the disconnect. I have seen 
this on many dymanics
installs through citrix. Ideally the dynamics client flips out and casues a 
session disconnect or crash.

I have always been able to fix this issues with the foolowing regkey setting on 
the citrix server:

HKEY_LOCAL_MACHINE\SOFTWARE\Citrix\CtxHook\AppInit_Dlls\Multiple Monitor Hook
Name: Flag
Typ: REG_DWORD
Value: 0
(Default is 4)

The above is for 32 bit, here is the 64 bit version:

HKEY_LOCAL_MACHINE\SOFTWARE\Wow6432Node\Citrix\CtxHook\AppInit_DLLs\Multiple
Monitor Hook

For those who have come across this key before, yes it's the old multimonitor 
hook issue. This turns the
MM hook globally so if you want to be specific and only turn it of for the app 
involved, do the following:

HKEY_LOCAL_MACHINE\SOFTWARE\Citrix\CtxHook\AppInit_Dlls\Multiple Monitor Hook
"Exclude"="dynamics.exe"

The "dynamics.exe" is the name of the dynamics client app. For those that use 
the program called
Navision on Citrix, you will know that this is a dynamics database with a front 
end of "finsql.exe" so the
entry for Navsion is:

HKEY_LOCAL_MACHINE\SOFTWARE\Citrix\CtxHook\AppInit_Dlls\Multiple Monitor Hook
"Exclude"="finsql.exe"

I don't remember if a reboot is required, but worth a reboot any way. Even 
though this fix is related to
multi-monitor setup, it still fixes one a sinlge monitor scenario. Read more 
here:

http://support.citrix.com/article/CTX110301

--
Warren Simondson

Ctrl-Alt-Del IT Consultancy Pty Ltd

Website: http://www.ctrl-alt-del.com.au








On Thu, Nov 4th, 2010 at 3:13 AM, "Landin, Mark" <Mark.Landin@xxxxxxxxxxxxxxxx> 
wrote:

> Running CPS 4.5 on Windows 2003 Standard, 32-bit, SP2. 4 GB RAM.
>
> Our primary application is Microsoft Dynamics GP as a published
> application. This has a memory footprint of 150-200MB per user, and we
> currently are seeing 20-25 users per box (i.e., 5GB+ of memory footprint)
>
> Recently our users have begun experiencing session disconnects, especially
> when coming back from being idle (walked away from their desk, turned to
> take a phone call, etc). We have no timeouts set on the ICA sessions and
> sometimes connection is lost even if they turn away for less than 5
> minutes.
>
> One theory is that the processes are dying because the servers are under
> memory stress (which they are), and that Windows is killing these
> processes to relieve the pressure. It may be killing the dynamics.exe
> process itself, or the ODBC connection between the application and the SQL
> server is being lost, which apparently causes the Dynamics GP application
> to wet itself and die.
>
> So, my question(s) is/are:
> -          When Windows is under severe memory pressure, will it just
> terminate processes? If so, how does it determine which ones?
> -          When it does so, will there be anything logged in Event Viewer?
> If not, how can one confirm that in fact Windows is killing processed in
> order to reduce memory footprint?
>
> If this is in fact the root cause, of course we have some options, like:
> -          build more servers (scale out)
> -          build bigger servers (scale up). We may be able to do 64-bit,
> but that will break new ground for us and I don't know how long that will
> take us to do. Would going to 2003 32-bit Enterprise Edition, and
> providing 8 or 16GB of RAM, help? Again, we are theorizing that the server
> is running out of RAM resources, not necessarily running out of kernel
> memory resources, so would throwing more RAM by way of Enterprise edition
> going to help?
>
> Thanks for the assistance.
>
> ________________________________
> This message and any attachments may be a confidential attorney-client
> communication or otherwise be privileged and confidential. If you are not
> the intended recipient, any review, distribution or copying of this
> transmittal is prohibited. If you have received this transmittal in error,
> please reply by e-mail and delete this message and all attachments
>
>

************************************************
For Archives, RSS, to Unsubscribe, Subscribe or
set Digest or Vacation mode use the below link:
//www.freelists.org/list/thin
Follow ThinList on Twitter
http://twitter.com/thinlist
************************************************


This message and any attachments may be a confidential attorney-client 
communication or otherwise be privileged and confidential. If you are not the 
intended recipient, any review, distribution or copying of this transmittal is 
prohibited. If you have received this transmittal in error, please reply by 
e-mail and delete this message and all attachments

************************************************
For Archives, RSS, to Unsubscribe, Subscribe or
set Digest or Vacation mode use the below link:
//www.freelists.org/list/thin
Follow ThinList on Twitter
http://twitter.com/thinlist
************************************************

Other related posts: