Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 13 Sep 2000 01:21:50 -0700
From:      "Crist J . Clark" <cjclark@reflexnet.net>
To:        JP <jr@paranoia.demon.nl>
Cc:        freebsd-questions@freebsd.org
Subject:   Re: Dual homed host routing problem
Message-ID:  <20000913012150.M69158@149.211.6.64.reflexcom.com>
In-Reply-To: <200009122019.WAA00707@pig.bigmama.xx>; from jr@paranoia.demon.nl on Tue, Sep 12, 2000 at 10:02:09PM %2B0200
References:  <20000911232915.J69158@149.211.6.64.reflexcom.com> <200009122019.WAA00707@pig.bigmama.xx>

next in thread | previous in thread | raw e-mail | index | archive | help
On Tue, Sep 12, 2000 at 10:02:09PM +0200, JP wrote:
> 
> 
> >>>>> On Mon, 11 Sep 2000, "Crist" == Crist J . Clark wrote:
> 
> 
>   Crist> The gateway_enable="YES" and manually setting
>   Crist> net.inet.ip.forwarding is redundant, that's what the
>   Crist> gateway_enable does.
> 
> Ok
> 
>   Crist> How have you tested it? What are the symptoms of yur problem?
>   Crist> Saying ed1 does not "fully forward" is a little unclear? Are
>   Crist> you trying to connect from a host on ed0's LAN to one on
>   Crist> ed1's? Or the other way around? Can the gateway connect to
>   Crist> machines on either LAN? -- Crist J. Clark cjclark@alum.mit.edu
> 
> 
> What happens is that from both the gateway machine and all hosts on the 
> 192.168.118 network, I am able to access the  ed1 interface, but 
> anything beyond the ed1 (other computers connected to that subnet, a 
> router that sits there, etc) are unreachable.  But they are not totally 
> unreachable, the traffic seems to get to the 172... address, but does 
> not get routed back.  I say that because the message returned is "host 
> is down" after only about 10 seconds

You should only get a 'host is down' message for machines on the local
net. I believe that is what you get after the ARP fails.

What I would still like to know is can the gateway machine itself
communicate with other machines on the 172.16.1.0/24 net? Can you ping
172.16.1.2 from the gateway, for example?

> but if i for example attempt to 
> ping a host that really does not exist, telnet will wait for many 
> minutes then the eventual message is "Unable to connect to remote 
> host".  Additionally, I see the traffic hit the nics/router  on the 
> other side (the lights flash).

Have you done tcpdump's on each interface to see what is getting in
and out?

> I also see now after trying Steve van den Akker's suggestion to use the 
> "cheat sheet" that natd is reporting "failed to write packet back (host 
> is down).
> 
> Any ideas what I am doing wrong?

NAT? You never said you were doing NAT. I suppose you have some
firewall rules too? So, here is what we know and still want to know,

  1) Can the gateway itself reach hosts on the 172.16.1.0/24 net? 
     Answer: ?

  2) Can the gateway itself reach hosts on the 198.168.118.0 net?
     Answer: Not explicitly said, but since the next one works the
             other direction I'll assume yes.

  3) Can the 198.168.118.0 hosts reach the gateway?
     Answer: No.

  4) Can the 172.16.1.0/24 hosts reach the gateway?
     Answer: ?

  5) Can 172.16.1.0/24 hosts reach the 198.168.118.0 net?
     Answer: ?, but if you are doing NAT (and unless you masked your
             address numbers, I don't know why you would be), this
             will be no.

  6) Can 198.168.118.0 hosts reach the 172.16.1.0/24 net?
     Answer: No.

  8) If (1), (2), (3), or (4) is "no," do a tcpdump on the gateway's
     interface of interest during the test and see if anything is going
     on.

  9) If (5) or (6) fails, do tcpdump's on _each_ interface while
     attempting the connections an see what happens.

> > On Mon, Sep 11, 2000 at 11:33:49PM +0200, JP (J. Patrick Russell) wrote:
> > > HELP!  I'm trying to setup a dual homed host with two ethernet devices.  Only 
> > > one device seems to be forwarding.  ed1 never seems to fully forward, I have 
> > > switched the card out with another type (rl0), but the behavior was identical 
> > > -- only interface ed0 worked.  ed1 is recognized and the cards do not seem to 
> > > conflict:
> > > 
> > > dmesg relevant output:
> > > 
> > > ed0: <NE2000 PCI Ethernet (RealTek 8029)> port 0x6000-0x601f irq 11 at device 
> > > 12.0 on pci0
> > > ed0: address 52:54:00:da:22:0c, type NE2000 (16 bit)
> > > ed1: <NE2000 PCI Ethernet (RealTek 8029)> port 0x6100-0x611f irq 10 at device 
> > > 14.0 on pci0
> > > ed1: address 52:54:00:da:21:6d, type NE2000 (16 bit) 
> > >
> > > and ed1 is automatically added to the routing table:
> > > 
> > > Internet:
> > > Destination        Gateway            Flags      Netif Expire
> > > default            172.16.1.2         UGSc        ed1
> > > 127.0.0.1          127.0.0.1          UH          lo0
> > > 172.16.1/24        link#2             UC          ed1 =>
> > > 172.16.1.2         link#2             UHLW        ed1 =>
> > > 172.16.1.3         52:54:0:da:21:6d   UHLW        lo0
> > > 192.168.118        link#1             UC          ed0 =>
> > > 192.168.118.1      52:54:0:da:22:c    UHLW        lo0
> > > 192.168.118.2      0:e0:4c:66:77:3d   UHLW        ed0    750
> > > 
> > > getting to the ed1 interface from the ed0 subnet works, but nothing gets 
> > > forwarded beyond the nic.
> > > gateway_enable="YES" is set in rc.conf, and sysctl is set so 
> > > net.inet.ip.forwarding=1
> 

-- 
Crist J. Clark                           cjclark@alum.mit.edu


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




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20000913012150.M69158>