From owner-freebsd-questions  Fri Apr 28  6:57: 7 2000
Delivered-To: freebsd-questions@freebsd.org
Received: from plains.NoDak.edu (plains.NoDak.edu [134.129.111.64])
	by hub.freebsd.org (Postfix) with ESMTP id AAED637B6E6
	for <freebsd-questions@FreeBSD.ORG>; Fri, 28 Apr 2000 06:57:05 -0700 (PDT)
	(envelope-from tinguely@plains.NoDak.edu)
Received: (from tinguely@localhost)
	by plains.NoDak.edu (8.9.3/8.9.3) id IAA03808;
	Fri, 28 Apr 2000 08:56:46 -0500 (CDT)
Date: Fri, 28 Apr 2000 08:56:46 -0500 (CDT)
From: Mark Tinguely <tinguely@plains.NoDak.edu>
Message-Id: <200004281356.IAA03808@plains.NoDak.edu>
To: freebsd-questions@FreeBSD.ORG, olive@deep-ocean.net
Subject: Re: looutput: mbuf allocation problem
Sender: owner-freebsd-questions@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.ORG


you must see allocation failures when you execute the command:

        $ netstat -m

With a MAXUSERS of 64, you should get 2560 NMBCLUSTERS (from the calculation
in sys/kern/uipc_mbuf.c). you can add more by the kernel option:

        options         NMBCLUSTERS=16384	# (or possibly 8192)

If you are just low of MBUF because of heavy traffic, this should do the
trick; if there is a MBUF leak in 4.0, then this will only postpone the
network hang.

I am surprised that the loopback output routine in FreeBSD 4.0 copies
every packet when only the KAME code (according to the comment) requires
the mbuf to be contiguous.

--mark tinguely.


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-questions" in the body of the message