Hello Sanford,

Thursday, January 6, 2005, 6:16:28 PM, you wrote:

>> I've  narrowed  it  down  to  too  many  TCP  sockets remaining in a
>> time_wait state during mail delivery.

SW> Clearing TIME_WAIT sockets more quickly:

SW> HKLM\SYSTEM\CurrentControlSet\Services\Tcpip\Parameters\TcpTimedWaitDelay

I read somewhere that was not a great way to manage the problem. But,
since I'm delivering very small datasets over LAN speeds then a
typical MSL of 2 minutes would be extremely high. But, I can't get any
definite handle on whether MSL is a hard figure or a function of mean
RTT. RFC793 defines maximum length that a socket pair should not be
used as 2 MSL, thus MS's setting of 4 minutes on the TIME_WAIT.

Any thoughts?

You know....this may explain the 8.14/7.13 anomaly. The 7.13 box is
about 25% underpowered related to the 8.14. It may be that it just
runs out of horsepower before it runs out of sockets.


SW> HKLM\SYSTEM\CurrentControlSet\Services\Tcpip\Parameters\MaxUserPort

Didn't know about this one. Will try it.

Still can't explain why 8.14 has no problem delivering to PF on BSD
4.07. Never see more than 30-40 sockets in TIME_WAIT.

-- 
Best regards,
 David                            mailto:[EMAIL PROTECTED]


To Unsubscribe: http://www.ipswitch.com/support/mailing-lists.html
List Archive: http://www.mail-archive.com/imail_forum%40list.ipswitch.com/
Knowledge Base/FAQ: http://www.ipswitch.com/support/IMail/

Reply via email to