Linux – Can’t add default route – RTNETLINK answers: Network is unreachable

linuxnetworkingrouting

On an Ubuntu 14.04 server (3.16.0-30) I'm trying to add a default route and getting error as follows:

ip route add default via 172.16.111.99 dev bond0.11
RTNETLINK answers: Network is unreachable

This box has address 192.168.11.23 on bond0.11. I have the following routes configured statically (edited down to relevant subset):

172.16.111.0/24 via 192.168.11.18 dev bond0.11 
192.168.11.0/24 dev bond0.11  proto kernel  scope link  src 192.168.11.23

I can ping 172.16.111.99 and here is traceroute's output:

traceroute to 172.16.111.99 (172.16.111.99), 30 hops max, 60 byte packets
 1  192.168.11.18  0.126 ms  0.096 ms  0.081 ms
 2  172.16.111.99  0.149 ms  0.142 ms  0.112 ms

Why is the default route addition being rejected when I clearly have access to that network/host? I'm obviously doing something wrong…

Best Answer

The issue is that the next hop or gateway in routes need to be on the same network as the one you are connected to.

i.e. you can set your default route to use 192.168.11.18. In turn that system will then know how to get to the 172.16.x.x network.

Related Solutions

Macos – Getting “no route to host”in Mac OS X 10.5.8

I'm guessing it's not related to ICMP being blocked. If that was the case, the response would simply timeout, not result in a "no route to host" message. "No route to host" means exactly that: when the destination of the ICMP message is not on the local subnet, it will be forwarded to the machine's default gateway for it to handle. If there is no default gateway, or the gateway has no information in its routing table to forward the packets, you'll see "no route to host."

Both your ping and traceroute commands are resolving the domain name to an IP address (albeit different addresses), so it appears that the DNS resolution is probably working okay. So I'd double check that your default gateways are the same/properly configured on the iMac first.

Also, from what I've seen, OS X seems to favor WiFi to a wired connection when both are connected. Check that you only have one connected (if you happen to be using both) and see if it works.

Linux – How to route specific network to a virtual machine

The error message indicates that the host you are trying to add this route on doesn't know where 192.168.0.2 is. You will need to provide that route first.

In general: if you want the 172.x.x.x network range to be accessible from the 10.x.x.x network, you will somehow need to let the computers in the 10.x.x.x network know that your host is the router for the 172.x.x.x range.

You will also need to set up adequate routing on the host machine for that range:

ip route add 172.x.x.x/xx via 192.168.0.2

If you have a split-horizon VPN, you will also need to advertise the 10.x.x.x network ranges (and possibly the 192.x.x.x ranges) to your VPN clients.

In order to debug your network routing, I suggest you get acquainted with the ip route get command. It displays you which interface a target address is seen through. Remember, routing needs to work both ways. The return packets must find their way back.

Best Answer

Related Solutions

Macos – Getting “no route to host”in Mac OS X 10.5.8

Linux – How to route specific network to a virtual machine

Related Question