Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 21 Jul 1995 17:34:25 +0100 (BST)
From:      Nik Clayton <Nik.Clayton@brunel.ac.uk>
To:        questions@freebsd.org
Subject:   Routing and/or sendmail problem
Message-ID:  <17546.9507211634@molnir.brunel.ac.uk>

next in thread | raw e-mail | index | archive | help
This is a general SOS. 

Situation
~~~~~~~~~
    I've got one machine with two interfaces:

        tun0, kernel tunnel device used with iijppp. My ISP allocates a
        static IP address, so I've put this in /etc/sysconfig. PPP works
        fine.

        ep0, 3C509b network card, configured on 0x390, IRQ 11. It's
        probed at boot up, gets recognised and configured properly.

     These interfaces are on two completely different networks. The tun0
     interface has the IP address 158.152.128.94 (surfs-up.demon.co.uk).
     The ep0 interface has the IP address 194.159.66.50
     (elbereth.blueberry.co.uk).

     My intention is to create a local network consisting of Macs
     running MacTCP, and this PC (all on the 194.159.66.0 net) and have
     the PC as a gateway so that these machines can connect to the
     Internet (using the tun0 interface).

What I've done
~~~~~~~~~~~~~~
     a) Compiled a kernel with "options GATEWAY" in the config file.

     b) Put the following lines in /etc/sysconfig

          hostname=elbereth.blueberry.co.uk
          .
          .
          .
          network_interfaces="lo0 tun0 ep0"
          ifconfig_lo0="inet localhost"
          ifconfig_tun0="inet surfs-up 158.152.1.222 netmask 0xffffffff"
          ifconfig_ep0="inet elbereth.blueberry.co.uk netmask 0xffffff00 link2"

     c) /etc/hosts looks like

          127.0.0.1             localhost
          194.159.66.50         elbereth.blueberry.co.uk        elbereth
          158.152.128.94        surfs-up.demon.co.uk            surfs-up

     d) /etc/resolv.conf looks like

          nameserver 158.152.1.193
          nameserver 158.152.1.65

        These are Demon's two nameservers. I'm not running 'named'.

     e) /etc/host.conf looks like

          hosts
          bind
   
     I'm not doing any routing myself. I'm relying on the kernel
     noticing the two interfaces and setting up a gateway between them
     (which, as far as I can tell, it's doing).

What happens
~~~~~~~~~~~~
     I really thought I'd managed to get this working. I log in, and the
     machine identifies itself as elbereth.blueberry.co.uk. I can 
     ``ping surfs-up.demon.co.uk'' and it pings correctly. I can run PPP
     and get to the outside world with things like Netscape, telnet and
     ftp. Almost everything is working perfectly.

     What's stopped worked is e-mail <grr>.

     Up until about four hours ago this machine only knew itself as
     "surfs-up.demon.co.uk". The ep0 interface was known to the kernel
     but I hadn't ifconfig'd it, and the machine was not on the 194.0
     net, only the 158.0 one. 'elbereth.blueberry.co.uk' didn't exist.

     At the time, e-mail to "nik@surfs-up.demon.co.uk" worked perfectly.
     As did e-mail to "nik@blueberry.co.uk" (my ISP forward's mail to
     the blueberry domain to the surfs-up domain).

     Ever since I made this change e-mail has been bouncing.

I think...
~~~~~~~~~~
     My ISP is re-writing mail to foo@surfs-up.demon.co.uk. It's
     connecting to the SMTP port on my machine expecting it to announce
     that it's connected to surfs-up.demon.co.uk. But my machine (even
     though the connection is on interface tun0) announces itself as
     elbereth.blueberry.co.uk.

     I can confirm this by doing

         telnet surfs-up.demon.co.uk smtp

     and get 

         Trying 158.152.128.94
         Connected to surfs-up.demon.co.uk.
         Escape character is '^]'.
         220-elbereth.blueberry.co.uk Sendmail 8.6.11/8.6.9 ready at
Fri, 21 Jul 1995 17:13:00 GMT 
         220 ESMTP spoken here

     So I think that what I'm asking is in reality a sendmail question.
     But I'm enough of a routing neophyte to not be 100% sure. Hence all
     the routing information above.

     I've had sendmail problems before, when I was running as 'surfs-up'
     but receiving mail addressed to 'blueberry'. This was fixed by
     putting the lines

         Cwsurfs-up.demon.co.uk
         Cwblueberry.co.uk

     in /etc/sendmail.cf.

     As is traditional at times like this, ``I haven't touched anything
     else!''. I haven't been near sendmail.cf in the past couple of
     weeks, and was receiving e-mail quite happily this morning (before
     I started playing around with changing machine names and the like).

Anyone got any suggestions? I've ploughed through "TCP/IP Network
Administration" and "Essential System Administration" both by O'Reilly
to no avail (other than to confirm that I think I've done the routing
stuff right), so any pointers would be greatly appreciated.

N
=-[Opinion, n: See the above text for an example]=-=[Kibo #: e]-[RYRYRY]=-=
=-[The Silly Sod Society: To perfect and to swerve]=-[beable]-=[TP U BG]=-=
Cleveland still lives.  God ____must be dead.



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?17546.9507211634>