SUMMARY: /etc/netmasks

Date: Mon Sep 28 1998 - 14:05:25 CDT

Thanks to:

Eugene Kramer <>
David Evans <>
Michael Wang <>
Taco Hettema <>
Jonathan Sturges <>
Frank Fiamingo <>
Brion Leary <>

*** Summary ***
Consensus is that there shouldn't be any problem in this scenario;
others are already doing this. Frank Fiamingo contributed the
following explanation:

> Solaris 2.5.1 uses BSD4.3 routing where the following rules are
> observed:
> 1. If the route is a subnet of a Class A/B/C network prefix and
> the machine has an interface using the same prefix, then the router
> assumes it should use the netmask from the local interface. And
> all interfaces on the machine with that same subnet prefix should
> use the same netmask.
> 2. If the route is a subnet of a Class A/B/C network prefix and the
> machine does NOT have an interface using the prefix, then the router
> assumes that it should use the default netmask for that class: A/8,
> B/16, C/24.

This basically means that on a locally-attached subnet, the netmask
of the corresponding interface is respected, whatever that may be.
If the subnet is _not_ local, then a standard mask is used. So as
long as it's local, a mask like is cool.

*** Original Post ***
OK, I'm pretty sure I understand what the limitations are for VLSM
on Solaris 2.5.1, i.e. the route command doesn't have a parm for a
netmask, so you're OK if it's locally attached.

My question is whether there's any problem using a mask such as in /etc/netmasks. I'm assuming that the netmask
is gathered in the /etc/rc2.d/S72inetsvc script, according to the
setting in /etc/nsswitch.conf for netmasks:. We're not using NIS,
and using only static routing across a quad ethernet card.

Derek Schatz Voice: 408-563-4198
Senior Systems Analyst Fax: 408-986-2822
Applied Materials, Inc. E-mail:

This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:12:49 CDT