Home / server / Questions / 991352
Accepted
Aleks G
Asked: 2019-11-12 04:11:48 +0800 CST

Routing doesn't work in linux system

  • 772

I'm struggling to get the networking working working on a linux VM. As is after the setup, I have this ifconfig:

# ifconfig
eth0: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1500
        inet 172.19.190.144  netmask 255.255.0.0  broadcast 172.19.255.255
        inet6 fe80::215:5dff:febe:6707  prefixlen 64  scopeid 0x20<link>
        ether 00:15:5d:be:67:07  txqueuelen 1000  (Ethernet)
        RX packets 30629  bytes 3428768 (3.2 MiB)
        RX errors 0  dropped 409  overruns 0  frame 0
        TX packets 351  bytes 53232 (51.9 KiB)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

lo: flags=73<UP,LOOPBACK,RUNNING>  mtu 65536
        inet 127.0.0.1  netmask 255.0.0.0
        inet6 ::1  prefixlen 128  scopeid 0x10<host>
        loop  txqueuelen 1000  (Local Loopback)
        RX packets 78  bytes 6600 (6.4 KiB)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 78  bytes 6600 (6.4 KiB)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

And this routing table:

# ip route
default via 172.19.190.1 dev eth0 proto static metric 100
172.19.0.0/16 dev eth0 proto kernel scope link src 172.19.190.144 metric 100

With this setup, I can ping the gateway and external internet hosts, but I cannot ping hosts on 172.19.180.x subnet, getting "host unreachable".

When I then manually do

# ip route add 172.19.0.0/16 via 172.19.190.1 dev eth0

Then everything works as expected. I now add this line into /etc/sysconfig/network-scripts/route-eth0:

172.19.0.0/16 via 172.19.190.1

After the reboot, my routing table looks like this:

# ip route
default via 172.19.190.1 dev eth0 proto static metric 100
172.19.0.0/16 dev eth0 proto kernel scope link src 172.19.190.144 metric 100
172.19.0.0/16 via 172.19.190.1 dev eth0 proto static metric 100

And I again cannot get to 172.19.180.x subnet. The difference between this and the manual addition with ip route add is that the last line has proto static metric 100 at the end, whereas with the manual way it doesn't.

What do I need to do to get everything working on a reboot?

UPDATE: here's my ifcfg-eth0 file:

BOOTPROTO=none
DEFROUTE=yes
IPV4_FAILURE_FATAL=no
IPV6INIT=no
IPV6_AUTOCONF=yes
IPV6_DEFROUTE=yes
IPV6_FAILURE_FATAL=no
IPV6_ADDR_GEN_MODE=stable-privacy
NAME=eth0
UUID=.... (masked)
DEVICE=eth0
ONBOOT=yes
IPADDR=172.19.190.144
PREFIX=16
GATEWAY=172.19.190.1
DNS1=172.19.190.240
DNS2=172.19.190.242

UPDATE2:

If I remove the second route (... src 172.19.190.144...) and only leave one default, then everything work correctly. Then the question becomes: where does the second route come from and how do I get rid of it?

linux

1 Answers

  1. Best Answer

    The problem is that your have two routes to 172.19/16. Since they have the same netmask length (16 bits) the second one overrides the third one.

    The third route:

    172.19.0.0/16 via 172.19.190.1 dev eth0 proto static metric 100

    was added by the /etc/sysconfig/network-scripts/route-eth0. The second one:

    172.19.0.0/16 dev eth0 proto kernel scope link src 172.19.190.144 metric 100

    has a proto kernel, so it was added by the script that configured the IP address of your interface.

    In the following I am guessing that your are using Fedora. The netmask/network prefix that your used to configure you IP address is too wide. You need to add to the file /etc/sysconfig/network-scripts/ifcfg-eth0 a line like:

    PREFIX=24

    Answer to update 2: When you add an IP address, say 1.2.3.4, to an interface, a couple of routes are added too. The exact routes depend on the tool:

    • ip addr will force you to specify the network prefix (size of the netmask) of 1.2.3.4 or complain vehemently. If you execute: ip addr add dev eth0 1.2.3.4/24, you' get a 1.2.3.4/24 dev eth0 route as bonus (+ 2 routes you don't see, because they are in the local routing table, cf. ip route show table local).

    • ifconfig will assume your network has a prefix of 24 if you do not specify a netmask and add a 1.2.3.4/24 dev eth0 route.

    So a direct answer to your question is: the PREFIX=16 line in /etc/sysconfig/network-scripts/ifcfg-eth0 caused the second route to appear.

    • 0