From owner-freebsd-hackers  Mon Apr  1 10:39:14 1996
Return-Path: owner-hackers
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.3/8.7.3) id KAA14763
          for hackers-outgoing; Mon, 1 Apr 1996 10:39:14 -0800 (PST)
Received: from brasil.moneng.mei.com (brasil.moneng.mei.com [151.186.109.160])
          by freefall.freebsd.org (8.7.3/8.7.3) with ESMTP id KAA14757
          for <hackers@freebsd.org>; Mon, 1 Apr 1996 10:39:12 -0800 (PST)
Received: (from jgreco@localhost) by brasil.moneng.mei.com (8.7.Beta.1/8.7.Beta.1) id MAA24760 for hackers@freebsd.org; Mon, 1 Apr 1996 12:38:40 -0600
From: Joe Greco <jgreco@brasil.moneng.mei.com>
Message-Id: <199604011838.MAA24760@brasil.moneng.mei.com>
Subject: SO_KEEPALIVE
To: hackers@freebsd.org
Date: Mon, 1 Apr 1996 12:38:39 -0600 (CST)
X-Mailer: ELM [version 2.4 PL24]
Content-Type: text
Sender: owner-hackers@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

Question, stupid, nonetheless a question:

Does anybody see any advantage to allowing for a way to set a systemwide
SO_KEEPALIVE default of ON?

My reason for asking:  from time to time, I run into situations where
software does not correctly terminate connections, most often due to broken
Portmasters or broken TCP/IP implementations.  For example,

smp       2983  0.0  0.0   412  348 ??  I    13Mar96    0:01.79 -ws1-6.gmttech.ods.net: smp: STOR test.zip                                  (ftpd)

or the dozens of hung nnrp processes that get into write() and never get out
on daily-planet.execpc.com.  My usual fix is to patch the source to enable
KEEPALIVE...

I don't mind a local kernel patch but it seems to me that many ISP's might 
find this useful.

... JG