On a server that is running Microsoft Forefront Unified Access Gateway (UAG) 2010, you may experience the following symptoms:
You received intermittent socket exhaustion alerts.
You notice high CPU usage and increased memory usage in the UAG IIS worker process (w3wp.exe).
You find that event ID 111 is logged in the Windows event log.
For example, consider the following scenario:
A web application uses an application-specific host name, or a Microsoft SharePoint application is configured as the initial internal application for the UAG 2010 trunk.
A request is made to the application for a fully qualified domain name (FQDN) that differs from the public host name that is configured in the application.
In this scenario, an internal loop may be encountered that leads to socket exhaustion and to high CPU usage and high memory usage during the loop. The loop does recover as soon as the socket exhaustion is encountered. Then, resources are released, and the working resource levels revert to typical levels.
To resolve this problem, install the service pack that is described in the following Microsoft Knowledge Base article:
2710791 Description of Service Pack 2 for Forefront Unified Access Gateway 2010
To work around this issue, set the default application of the trunk to the portal application.
Microsoft has confirmed that this is a problem in the Microsoft products that are listed in the "Applies to" section.
To enable this fix, follow these steps:
Create the following registry subkey:
DWORD name: RejectUnknownHost
Value data: 1
Start the UAG configuration.
Run IISReset as an administrator to restart Internet Information Services (IIS).
For more information about software update terminology, click the following article number to view the article in the Microsoft Knowledge Base:
824684 Description of the standard terminology that is used to describe Microsoft software updates