Traceroute uses incorrect interface to get out

[wolfej]

I am running Solaris 2.6 (SunOS 5.6) on a box with two nic cards called hme0 and hme1. Hme0 is configured with 141.114.141.183 (with a 255.255.0.0 subnet mask) and hme1 has a private address of 10.5.2.2 with a 255.255.255.0 netmask.

The story is somewhat complicated regarding recent changes that have been made, so I will try to leave out some of the details and ask only the salient questions!

The default gateway is setup as a 141.114.x.x address (and my understanding is that I can only have one default gateway even though I am connected to two networks.)

The problem is this: I am unable to reach (ping) the private address space's default gateway of 10.5.2.1 and it is plugged into exactly the same switch as the solaris host. When I try to see what path it is taking by using a traceroute I get this message from the os: "traceroute: warning multiple interfaces found; using 141.114.141.183@hme0."

When I try "route add 10.5.2.0 10.5.2.2" I get an error that the network is unreachable!

So, it appears that when I ping 10.5.2.1, the os will not use the correct nic to get out on the proper network. Is there a way I can force the os to use the correct nic for this path and at the same time to get it to recognize that the 10.5.2.x network IS indeed reachable?

 

[Annihilannic]

Please post the output of netstat -rn so we can see the routes. Note that you can use the -i switch on traceroute to specify which interface to use. No routes should be necessary to reach your gateway on that network... (as I think you realise).

Annihilannic.

 

[kociant]

You can manually add a route on the server as follows:

route add net 10 10.5.2.1 1

This will route the entire 10.x.x.x to the gateway address 10.5.2.1. You can also add a file in the /etc/rc2.d/ (for example) directory in a file (say S99static.routes). This will be read when the server is (re)booted.

 

[wolfej]

Thank you for your help guys.

The situation is so bizzare right now that I think I will let it rest. In short, the problem began when on the public network side, we needed to connect to a node on another (an additional) 10.x.x.x address space. I have added a route to that particular node and it seems to be talking to both 10 networks, but only as long as the interface itself is configured with a 255.0.0.0 netmask.

So it is working and I think I will leave it all alone as long as it continues to do so. If things should change again that should cause my boxes to stop connecting, I will consider changing my private address space to something like a 192.168.x.x.

Thanks for your help.