In addition, there is a Microsoft Tech Note stating this registry entry is *NOT* present to the registry , and the default value is 2 hours. This explains why the failure point of the delay occurred between 1 and 2 hours of being idle.
http://technet2.microsoft.com/windowsserver/en/library/af2e0d81-50cc-430 d-80e1-a2ccebfc68f21033.mspx?mfr=true KeepAliveTime Updated: March 28, 2003 HKLM\SYSTEM\CurrentControlSet\Services\Tcpip\Parameters Data type Range Default value REG_DWORD 0x1-0xFFFFFFFF (milliseconds) 0x6DDD00 (7,200,000 milliseconds = 2 hours) Description Specifies how often TCP sends keep-alive transmissions. TCP sends keep-alive transmissions to verify that an idle connection is still active. This entry is used when the remote system is responding to TCP. Otherwise, the interval between transmissions is determined by the value of the Parameters\KeepAliveInterval <http://technet2.microsoft.com/WindowsServer/en/library/734570a2-06d6-45 0e-b765-ccfa7530af491033.mspx> entry. By default, keep-alive transmissions are not sent. The TCP keep-alive feature must be enabled by a program (such as Telnet), or by an Internet browser (such as Internet Explorer). Note * Windows Server 2003 does not add this entry to the registry. You can add it by using the registry editor Regedit.exe. From: Action Request System discussion list(ARSList) [mailto:[EMAIL PROTECTED] On Behalf Of Tony Worthington Sent: Wednesday, January 09, 2008 12:47 PM To: arslist@ARSLIST.ORG Subject: Re: Delays after user tool sits idle ** Wow... Matt. Thanks for the information! I'm going to have to pass this along to our network folks. This might also fix our issues with our load-balancers not being able to track sessions that aren't persistent with a keep-alive (Remedy.) Thanks again! Tony -- Tony Worthington Sr. Technical Analyst Kohl's Department Stores [EMAIL PROTECTED] 262-703-5911 Matt Reinfeldt <[EMAIL PROTECTED]> Sent by: "Action Request System discussion list(ARSList)" <arslist@ARSLIST.ORG> 01/09/2008 11:04 AM Please respond to arslist@ARSLIST.ORG To arslist@ARSLIST.ORG cc Subject Re: Delays after user tool sits idle ** Tony, Bryan Waters and I have been working with the escalation engineers on this, and we've seen the following fix resolve the issue: Apply the below Registry settings, if in a Windows environment, to each Windows server that hosts an AR Server. If the AR Server is on Linux/UNIX, see this note: "It is depend on the server platform. If it is Windows, I think the steps are same for server and client. If it is Linux, run echo "300" >/proc/sys/net/ipv4/tcp_keepalive_time (default is 7200, 2 hours) To confirm it works, use WireShark for Windows and tcpdump for Linux to monitor the connection between client and server. On every keepalive_time, server should send out an acknowledge packet to client and client sends reply back. In this way, connection is alive for another keepalive_time interval." Open your registry and find the key below. Create new DWORD value named "KeepAliveTime" and set it to equal the number of milliseconds to wait before sending keep alive packets (the default is 2 hours - 7,200,000 milliseconds). Also create a new DWORD value called "KeepAliveInterval" and set it to equal the time in milliseconds between retransmissions of keepalives, once the KeepAliveTime has expired (the default is 1 second - 1000 milliseconds). Restart Windows for the change to take effect. Registry Settings System Key: <http://www.pctools.com/guides/help/registry-settings.php#system_key> [HKEY_LOCAL_MACHINE\System\CurrentControlSet\Services\Tcpip\Parameters] Value Name: <http://www.pctools.com/guides/help/registry-settings.php#value_name> KeepAliveTime, KeepAliveInterval Data Type: <http://www.pctools.com/guides/help/registry-settings.php#data_type> REG_DWORD (DWORD Value) The KeepAliveTime would be set to 300000 for our test (5 minutes). If the Registry changes resolve the issue for you, we could export the settings from the dev server's Registry and simply import them to the production machines before rebooting them. KeepAliveTime set to 300000 KeepAliveInterval set to 5000 Anyway, long story short, it appears that some network device is killing the idle TCP connection after n-minutes (value depends on the device settings). So, the solution is to send a keep alive signal from the server more often than the default 2 hours, or identify the device causing the problems and change its settings. Hope that helps you with your issue. BTW: we've requested that a KB get written up for this, but I don't know when/if that will happen. Matt Reinfeldt ________________________________ From: Action Request System discussion list(ARSList) [mailto:[EMAIL PROTECTED] On Behalf Of Tony Worthington Sent: Thursday, September 27, 2007 6:05 AM To: arslist@ARSLIST.ORG Subject: Re: Delays after user tool sits idle ** I think that's a good idea. ethereal+tcpdump should spit sometime out. At least more than the user tool logs. :-) I'll let everyone know what happens. Still no response from support.. -tony -- Tony Worthington Sr. Technical Analyst Kohl's Department Stores [EMAIL PROTECTED] 262-703-5911 __Platinum Sponsor: www.rmsportal.com ARSlist: "Where the Answers Are" html___ __Platinum Sponsor: www.rmsportal.com ARSlist: "Where the Answers Are" html___ This message is subject to and does not create or vary any contractual relationship between TuringSMI, SMI Technologies, SMI Telco, its subsidiaries or affiliates and you. Internet communications are not secure and therefore the TuringSMI Group does not accept any legal responsibility for the contents of this message. Any views or opinions expressed are those of the author. This message is intended for the addressee ('s) only and its contents and any attached files are strictly confidential. If you have received it in error, please contact the sender on the number above. _______________________________________________________________________________ UNSUBSCRIBE or access ARSlist Archives at www.arslist.org Platinum Sponsor: www.rmsportal.com ARSlist: "Where the Answers Are"
