FAST Search crawling performance has been dropped in multi server farm

SharePoint 2010 and FAST 14 search integration to provide enterprise search solution.

As a administrator, we observed various crawling and performance related issues and found a common fix to most of issues\warnings\errors on FAST servers:

1. An unhandled exception occurred: The underlying connection was closed: An unexpected error occurred on a receive. at System.Net.HttpWebRequest.GetResponse() at Microsoft.SharePoint.Search.Extended.QRServer.ProxyBase.ProcessRequest(UriBuilder uriBuilder, String& contentType, Int64& contentLength) at Microsoft.SharePoint.Search.Extended.QRServer.ProxyWcf.InternalProcessRequest(String request) 
2. timed out while processing batch. 13141 - 13240. Removing processor server and notifying client of batch failure 
3. Failed to renew lease 
4. configuration: No response from the config server. Will wait 300 seconds for config server to respond before proceeding. 
5. JDBCConnector: Thread : Publisher0 has been inactive for more than 300 seconds. 

Solution

===========

- Fix most of above communication related errors\warning by making below changes on FAST server:

• To disable TCP/IP Chimney Offload
• To disable RSS
• Ensure all of the “offload” settings on the network adapter’s driver are set to Disabled

For more info:http://support.microsoft.com/kb/951037 

http://blogs.msdn.com/b/scottos/archive/2010/09/29/ensuring-a-successful-fast-search-server-2010-installation.aspx 

Use netsh int tcp show global command to check status of it: