none
2008 R2, IIS 7 and UNC problems, managed pipeline modes ? SMB Settings on UNC shares ?

    Question

  • Hello,
    I am posting in this forum, as recommended to me by a Support Engineer at Microsoft.
    I've copied over my original posting on the IIS.net forums as there have been some replies, but no solutions at this point
    http://forums.iis.net/t/1162209.aspx

    I am currently experiencing an interesting problem with IIS 7 on w2k8 r2 and UNC content in a virtual directory.

    We have a front end web server running IIS7.5 off a W2K8 R2 box, on Active Directoty. The application of interest is an ASP.NET 2.0 web application sourced from the local c:\ disk. Within this WebSite Application, we have a virtual folder which is mapped to a UNC file share on a back end server.

    This file server is running 2003 R2 server and is on the same domain.

    The UNC permissions are setup properly, and on the front end web application, the Application Pool is running under a user which has the proper permissions to source the content.

    By defaul, the application pool's Managed Pipeline mode is set to integrated.

    This works for a number of requests(not sure on exact number, but probably under 50), after which point, we start loosing the ability to source content from the file server (images, media, etc...) as mapped through the Virtual Folder (ie from the UNC share).

    We get the following ERROR in the Event Log on the W2K8 Front end server
    Failed to start monitoring changes to '\\FILESERVER\CONTENTSHARE\YYYYYY' because the network BIOS command limit has been reached. For more information on this error, please refer to Microsoft knowledge base article 810886. Hosting on a UNC share is not supported for the Windows XP Platform.

    Now the interesting bit, in trying to debug this further, if we change the Managed Pipeline mode of the application Pool to "Classic", I stop experiencing the problem (though I have not had a chance to test this extensively, but do not seem to be having the problem, no longer problem content, nor any error messages in the event log).

    In referring to the error message, I know there are a number of articles describing registry key changes which would allow for MaxCmds to be higher, however these typically describe changes to both the back end server, as well as the front end web server.
    http://learn.iis.net/page.aspx/198/smb-commands-registry-keys/
    http://weblogs.asp.net/hosterposter/archive/2006/10/30/Hosting-IIS-with-UNC-content-_2D00_-Network-BIOS-commands-and-other-errors.aspx

    Now the question, is WHY is this only happening in the integrated mode, and not in classic and how can we remedy this.
    The backend file server cannot easily be tweaked, as is is currently serving content to a production server runnning on W2003 IIS6 with the same application (same domain user defined in the app pools on both front end servers).

    Can someone tell me wheter we should schedule maintenance to do the registry tweaks on the back end file server, and whether this would change anything for the front end server running W2K8 R2.
    Why the difference in behavior between Classic VS Integrated Managed Pipeline Mode in accessing the UNC content.
    Sorry for cross posting as I've put this message up in microsoft.public.inetserver.iis but think the community might be more active on the forums site here.

    Any insight would be appreciated.

    ---
    To be clear, the same testing sequence on the 2008 server on the front end will stall within 50 or so requests when the Application Pool is set to use the Integrated mode in teh Managed pipeline, wehreas when running in classic mode, I do not seem to be able to hangup the application at all.

    We need to know if this is related to the back end SMB modes and the UNC shares on the 2003 server or why and how to we avoid the access issues when running the 2008 web application in "integrated mode" as we would prefer this native/default mode.


    Sincerely,
    Nicolai

    Tuesday, November 3, 2009 8:00 PM