— modern ops stuff —
Building a Router With a Solaris Zone
06 April 2011 // Solaris

I needed to integrate into my 192.168.1.0/24 home network, a machine whose ALOM I knew to be on 10.10.8.0/24. I don’t have any fancy routing hardware, so I thought it would be a good idea to try it in Solaris. I’m using SXCE build 130, but all this will work on OpenSolaris, Solaris Express, or anything based on either of those. All you need is Crossbow, because it’s based around VNICs. If you aren’t sure whether or not your system supports VNICs, run

# dladm 2>&1 | grep create-vnic

And if you get output, it does.

The plan is to create a dedicated router zone which forwards traffic between 192.168.1.0 and 10.10.8.0. I only have one physical NIC (atge0) in my workstation though.

I’m going to create a VNIC on top of atge0 specifically for my routing zone. It’ll be the 10.10.8.0 address. Because VNICs are free, I’ll also create one for 192.168.1.0

[global]
# dladm create-vnic -l atge0 vnic_rt0
# dladm create-vnic -l atge0 vnic_rt1

Now create the routing zone itself. Its IP addresses will be 10.10.8.1 on vnic_rt0 and 192.168.1.253 on vnic_rt1. I’m going to use my s-zone.sh script, which can do most things automatically

[global]
# print "10.10.8.0   255.255.255.0" >>/etc/netmasks
# ./s\-zone.sh create -e vnic_rt0=10.10.8.1,vnic_rt1=192.168.1.253 \
  -R rpool/zoneroot -D space/zonedata tap-router

Now, let’s have a look at the zone.

[global]
# zlogin tap-router
[tap-router]
# ifconfig -a
vnic_rt0:1: flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 4
      inet 10.10.8.1 netmask ffffff00 broadcast 10.10.8.255
vnic_rt1:1: flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 5
      inet 192.168.1.253 netmask ffffff00 broadcast 192.168.1.255

Looks good. Let’s see if we can ping the ALOM, which I happen to know is on 10.10.8.7.

# ping 10.10.8.7
10.10.8.7 is alive

And can we ping my workstation?

# ping tap-ws
tap-ws is alive

To make sure the zone comes up properly, that is that the VNICs are plumbed, you’ve got to put /etc/hostname.nic entries in to the zone

# print tapz-router >/etc/hostname.vnic_rt0
# print >/etc/hostname.vnic_rt1

Now we just have to tell Solaris to forward traffic between the two VNICs. Back in the day, this was a pain, messing about with ndd, but modern Solaris gives you lots of ways to do it, via ifconfig, routeadm and svcadm. I prefer routeadm

# routeadm
              Configuration   Current              Current
                     Option   Configuration        System State
---------------------------------------------------------------
               IPv4 routing   disabled             disabled
               IPv6 routing   disabled             disabled
            IPv4 forwarding   disabled             disabled
            IPv6 forwarding   disabled             disabled

           Routing services   "route:default ripng:default"

Routing daemons:

                      STATE   FMRI
                   disabled   svc:/network/routing/ripng:default
                   disabled   svc:/network/routing/route:default
                   disabled   svc:/network/routing/rdisc:default
                   disabled   svc:/network/routing/legacy-routing:ipv4
                   disabled   svc:/network/routing/legacy-routing:ipv6
                   disabled   svc:/network/routing/ndp:default

No routing there. You have to enable routing with the -e option, then update the routing configuration with -u. I’m not interested in IPV6. (Is anybody? We should be.)

# routeadm -ue ipv4-forwarding

-u makes the change effective immediately, -e enables.

Now I have to tell my workstation to use 192.168.1.253 to access the 10.10.8.0 subnet.

[tap-ws]
# route -p add net 10.10.8.0 192.168.1.253

The -p flag makes the route persistent. This is another Solaris 10 feature which saves us having to use custom scripts or services to set routes.

And that’s it. All done.

Tags: