0

We are running 11 nodes on GKE and get this error between a few hundred and a thousand times a day:

Runtime network not ready: Network plugin is not ready: Kubenet does not have netConfig. This is most likely due to lack of PodCIDR

I couldn't find much info related to the issue. All other posts I've found are fruitless.

There is this issue in Azure github that indicates people have given up w/o a resolution.

There is this inactive github issue as well.

And actually another SO question which has no resolution either.

The nodes are running alright, we are running pods on them w/o problems. But the nodes do get restarted randomly from time to time (not at the same time), so I am wondering whether these errors are related. And I'm trying to get some deeper insight on the issue.

Milkncookiez
  • 203
  • 2
  • 8

1 Answers1

1

This means that you choose a CIDR with less IPs than you need, when we deploy a cluster on Kubernetes be on Google Cloud or Baremetal we define a range of CIDR, you should commonly use a /16 CIDR like 10.50.0.0/16 this will give you options to use more than 65K ips internally, please remember that on Kubernetes you have IPs not only for pods but for services and other things as well, never choose less than /24 (~250 IPs) unless you will deploy a real little cluster.

Alex
  • 174
  • 3