0
votes

I have configured one kube master server and one kube minion server. both are on the CentOS 7 on master server, kube-scheduler , kube-controller-manager and kube-apiservr services are running along with docker and flanneldured etcd server is configured on the kube master server which is up and running properly. however, while configuring flanneld on minion server, getting error like:

"failed to retrieve network config : client : etcd clustr is unavailable or misconfigured"

i have verified already following 1: kube minion server and kube master server are able to ping by hostname 2: firewall is stopped on both the server 3: /etc/sysconfig/flanned file is configured to point to etcd host IP running on kube master

can someone please suggest why this error is coming.. what else needs to be checked on minion server ?

Regards

1
Have you tried accessing the Etcd service from the minion? Try something like :curl -L http://<etcd_host_ip>:<etcd_port>/version from the minion to check that it can reach the service.rwehner

1 Answers

0
votes

On each Kubernetes node instance, in /etc/sysconfig/flanneld you have to provide :

  • --etcd-endpoints
  • --etcd-prefix

And before that, did you push any network configuration in an etcd key ?

On my servers for example :

[root@jm-fou ~]# etcdctl --endpoints http://master:5001 ls /coreos.com/network /coreos.com/network/config /coreos.com/network/subnets

and

[root@jm-fou ~]# etcdctl --endpoints http://master:5001 get /coreos.com/network/config {"Backend": {"Port":8285,"Type":"udp"}, "Network":"172.29.0/16", "SubnetLen":24}

this json has been pushed by the folloging command :

etcdctl --endpoints http://master:5001 set /coreos.com/network/config {"Backend": {"Port":8285,"Type":"udp"}, "Network": "172.29.0/16", "SubnetLen":24}

and the kube-controller-manager is started specifying this network :

--cluster-cidr=172.29.0.0/16