From owner-freebsd-isp  Fri Mar 22 09:17:09 1996
Return-Path: owner-isp
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.3/8.7.3) id JAA15977
          for isp-outgoing; Fri, 22 Mar 1996 09:17:09 -0800 (PST)
Received: from kitten.mcs.com (Kitten.mcs.com [192.160.127.90])
          by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id JAA15969
          Fri, 22 Mar 1996 09:17:00 -0800 (PST)
Received: from mailbox.mcs.com (Mailbox.mcs.com [192.160.127.87]) by kitten.mcs.com (8.6.12/8.6.9) with SMTP id LAA21575; Fri, 22 Mar 1996 11:16:57 -0600
Received: by mailbox.mcs.com (/\==/\ Smail3.1.28.1 #28.5)
	id <m0u0ASK-000AvAC@mailbox.mcs.com>; Fri, 22 Mar 96 11:16 CST
Received: by mercury.mcs.com (/\==/\ Smail3.1.28.1 #28.5)
	id <m0u0ASI-0003ktC@mercury.mcs.com>; Fri, 22 Mar 96 11:16 CST
Message-Id: <m0u0ASI-0003ktC@mercury.mcs.com>
From: sigma@mcs.com (Kevin Martin)
Subject: RTM_LOSING "Kernel suspects partitioning"
To: freebsd-isp@freebsd.org
Date: Fri, 22 Mar 1996 11:16:49 -0600 (CST)
Cc: freebsd-questions@freebsd.org
Content-Type: text
Sender: owner-isp@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk


I have FreeBSD running on a Pentium-120, 128Mb RAM, SMC NIC (de0), with
over two hundred virtual domains.  A while ago, we were having unusual
network storms, and I did lots of experimentation with the virtual domain
set up to see if it was related.  In the end, I changed so many things
(eliminating the specific "route add" commands, for example), that I'm not
sure exactly what has cleared up the problem.

But in any case, I saw at the time, and have continued to see, this message
on "route -n monitor":

RTM_LOSING: Kernel Suspects Partitioning: len 124, pid: 0, seq 0, errno 0,
flags:<UP,GATEWAY,HOST,DONE>
locks:  inits: 
sockaddrs: <DST,GATEWAY>

Followed by IP addresses of the remote and the local host.  This seems to
happen for *any* connection to the machines, so there are a lot of these
messages.

I looked through the source, a few books, and on the net, but I can't find
the meaning of this.  Does it represent a problem?  It seems to.

I think the thing that actually fixed the network hangs was increasing
NMBCLUSTERS to 4096; there were occasional "mb_map full" messages before
that.

Please reply by e-mail, I'll summarize later.

Thanks,
Kevin Martin
sigma@mcs.com