Pods scheduled to a node receive an IP address from the node's Pod CIDR range. At this point pods can not communicate with other pods running on different nodes due to missing network routes.
In this lab you will create a route for each worker node that maps the node's Pod CIDR range to the node's internal IP address.
There are other ways to implement the Kubernetes networking model.
On each worker node, add the following routes:
WARNING: don't add the route associated to the POD CIDR for the current node (ex: don't add the 10.200.0.0/24 route if you are on the worker-0 node).
ip route add 10.200.0.0/24 via 172.16.0.20 # Don't add on worker-0
ip route add 10.200.1.0/24 via 172.16.0.21 # Don't add on worker-1
ip route add 10.200.2.0/24 via 172.16.0.22 # Don't add on worker-2
Don't take care of the
RTNETLINK answers: File exists
message, it appears just when you try to add an existing route, not a real problem.
List the routes in the kubernetes-the-hard-way
VPC network:
ip route
Output (example for worker-0):
default via 172.16.0.1 dev ens19 proto static
10.200.1.0/24 via 172.16.0.21 dev ens19
10.200.2.0/24 via 172.16.0.22 dev ens19
172.16.0.0/24 dev ens19 proto kernel scope link src 172.16.0.20
To make it persistent (if reboot), you need to edit your network configuration (depends on your Linux distribution).
Example for Ubuntu 18.04 and higher:
vi /etc/netplan/00-installer-config.yaml
Content (example for worker-0, don't specify the POD CIDR associated with the current node):
network:
version: 2
ethernets:
ens18:
dhcp4: true
dhcp4-overrides:
use-routes: false
ens19:
dhcp4: no
addresses:
- 172.16.0.20/24
gateway4: 172.16.0.1
nameservers:
addresses:
- 8.8.8.8
- 8.8.4.4
routes:
- to: 172.16.0.0/24
via: 172.16.0.1
scope: link
- to: 10.200.1.0/24
via: 172.16.0.21
- to: 10.200.2.0/24
via: 172.16.0.22