[102] in linux-net channel archive
NCSA httpd children hanging around in CLOSE
daemon@ATHENA.MIT.EDU (mbrennen@maverick.intecom.com)
Fri Feb 24 19:04:25 1995
From: mbrennen@maverick.intecom.com
Date: Fri, 24 Feb 95 15:18:40 PST
To: linux-net@vger.rutgers.edu
I'm not part of this list, so please respond directly. Al Longyear suggested that I
float this problem by this list, as he knew of a previous socket lingering close problem
in the kernel.
BTW, what is the proper submission to join the list?
I'm running NCSA httpd 1.3 on two boxes, one enet connected, one pppd connected. Both
are running 1.1.88; I am in the process of upgrading to 1.1.94, but I'm not there yet.
The problem is that I see lingering httpd children that netstat reports as being in CLOSE
state with a send_q=1; they hang around a short time (~ 2 minutes??) and disappear. I
see this very little with the enet box, but I see it quite a bit with the pppd connected
box. Not every child does this, but many do.
Is this a recognized problem, and if so what is the workaround / upgrade for it? I saw
this some in 1.1.75 also, the last kernel I ran heavily.
Thanks in advance -
Michael
mbrennen@fni.com
---------------
Michael Brennen Intecom, Inc.
mbrennen@intecom.com 5057 Keller Springs
(214) 447-8074 Dallas, TX 75248
