ip route

Every packet leaves by the most specific route that matches its destination; default matches everything else. ip route get ADDRESS is the most useful form in troubleshooting: it does the lookup for you and prints the next hop, the card and the source address the kernel would use. A route via a gateway must point at an address on one of your own networks.

ip route command usage

ip [-6] route [show]
ip route get ADDRESS
ip route add|replace|del PREFIX via GATEWAY [dev CARD] [metric N]
ip route help

Options

help

print every form of the command and the words a route takes, and stop

get 10.0.2.10

which route, next hop and card a packet to that address takes

add PREFIX via GW

send PREFIX to the neighbour GW

replace

add, or change the route if it already exists

metric 100

the route's cost: of two routes for the same prefix, the lower metric is used

del PREFIX via GW

delete a route; via says which, when the prefix has several

show default

only the default route

-6

the IPv6 table

ip route examples

ip route

Prints the routing table: one line per destination network and where it goes.

ip route get 10.0.2.10

Shows which next hop, card and source address a packet to 10.0.2.10 would use.

ip route show default

Shows only the default route, the gateway for everything not listed.

ip route add default via 10.0.1.2

Sends every packet with no better route to the router at 10.0.1.2.

ip route add 10.0.3.0/24 via 10.0.12.2

Sends packets for 10.0.3.0/24 to the next router, 10.0.12.2.

ip route del 10.0.3.0/24

Removes the route for 10.0.3.0/24.

ip -6 route

Prints the IPv6 routing table.

ip route add 10.0.3.0/24 via 10.0.12.2 dev eth1

Adds a route and names the card, for a next hop reachable on more than one.

ip route replace default via 10.0.1.3

Changes the default gateway in one step, adding the route if there was none.

ip route add 10.0.3.0/24 via 10.0.13.2 metric 200

Adds a backup route: the lower metric wins while both exist.

ip route add default nexthop via 10.0.1.2 weight 1 nexthop via 10.0.1.3 weight 1

Spreads traffic over two gateways (ECMP), per flow.

ip route add blackhole 192.0.2.0/24

Drops every packet for 192.0.2.0/24 without an answer.

ip route add 10.0.3.0/24 via 10.0.12.2 table 100

Adds the route to table 100, used only for packets that an ip rule sends there.

ip route show table all

Prints every table, local and policy ones included, not only main.

ip route show proto bgp

Lists only the routes a BGP daemon installed: proto says who added a route (kernel, static, dhcp, ospf, bgp).

Reading the output

$ ip route; ip route get 10.0.2.10
default via 10.0.1.2 dev eth0
10.0.1.0/24 dev eth0 proto kernel scope link src 10.0.1.10
10.0.2.10 via 10.0.1.2 dev eth0 src 10.0.1.10 uid 0
    cache

• default via 10.0.1.2: anything not matched below goes to the router at 10.0.1.2. No such line: only your own network is reachable.
• 10.0.1.0/24 dev eth0 proto kernel scope link: your own network, reached directly. The kernel adds it when the address is set.
• ip route get answers via 10.0.1.2 dev eth0 src 10.0.1.10: the next hop, the card, and the source address replies will come back to.

Pitfalls

• A router needs routes back as well: the reply to 10.0.1.10 is routed from the far side, by other tables.
• Nexthop has invalid gateway: the via address is not on any of your own networks.
• On a router running OSPF or BGP, routes also come from the daemon: see them with vtysh -c 'show ip route'.

Run it on real machines