SUMMARY: multihomed Sol8 / Apache / Incorrect IP no. use

From: Geoff Lane <zzassgl_at_twirl.mcc.ac.uk>
Date: Mon Aug 20 2001 - 03:56:48 EDT
I received no replies but some experimentation has led me to what appears to
have been the trivial problem that caused the situation described below.

For some reason the netmask for the hme0:1,:2,:3,:4 etc interfaces was set
to 255.255.0.0 instead of the correct value of 255.255.255.0 (ip_forwarding
was always set to 0, something I should have mentioned)

When I re-created the interfaces with
	# ifconfig hme0:1 plumb
	# ifconfig hme0:1 xxx.xxx.xxx.xxx netmask 255.255.255.0 up

the interface was activated and the system responded correctly to packets
directed to xxx.xxx.xxx.xxx.

I guess the OS saw a more general route out of the machine available on the
new multihome interfaces and decided to use it.

------------------

> I want to set up a Solaris8 web server to run a number of IP based apache
> web servers.  Each web server will have a unique IP number associated with
> it and be configured using the Listen option to respond only to requests
> directed at it's associated IP number.
> 
> The hardware has a single network interface and will be multihomed in the
> usual manner (ie hostname.hme:1 etc) The server has a primary IP and DNS
> entry that's not used for the web service.
> 
> I've done some experiments and although the multihoming and apache works
> fine there is one showstopper problem I'm not sure how to fix.
> 
> When the system is originating a network connection (ie sending syslog
> records or printer requests to a remote host) it's chosing to use the web
> server IP numbers rather than it's primary IP number, so the remote hosts
> see requests from unexpected machines and either incorrectly label the
> syslog messages or reject the print requests.
> 
> How can I instruct the system to only use it's primary IP number when
> orginating network connections yet also use the appropriate web server IP
> numbers when responding to web requests?
> 
> From what I've read, ip_strict_dst_multihoming, may be part of the answer
> but I'm not sure if it's appropriate in this case.

-- 
/\ Geoff. Lane. /\ Manchester Computing /\ Manchester /\ M13 9PL /\ England /\
Received on Mon Aug 20 08:56:48 2001

This archive was generated by hypermail 2.1.8 : Wed Mar 23 2016 - 16:25:02 EDT