All in one 部署k8s常见问题

All in one 安装k8s

kubeasz/quickStart.md at master · easzlab/kubeasz · GitHub

设置本机免密

将自己公钥放在authorized_keys

root@master1:/etc/kubeasz# journalctl -u kubelet | grep failed

Apr 17 20:08:36 master1 kubelet[84274]: I0417 20:08:36.824278   84274 nvidia.go:61] NVIDIA setup failed: no NVIDIA devices found

Apr 17 20:08:36 master1 kubelet[84274]: E0417 20:08:36.906828   84274 server.go:292] "Failed to run kubelet" err="failed to run Kubelet: misconfiguration: kubelet cgroup driver: \"systemd\" is different from docker cgroup driver: \"cgroupfs\""

Apr 17 20:08:42 master1 kubelet[84813]: I0417 20:08:42.277033   84813 nvidia.go:61] NVIDIA setup failed: no NVIDIA devices found

Apr 17 20:08:42 master1 kubelet[84813]: E0417 20:08:42.323635   84813 server.go:292] "Failed to run kubelet" err="failed to run Kubelet: misconfiguration: kubelet cgroup driver: \"systemd\" is different from docker cgroup driver: \"cgroupfs\""

解决：

docker中将资源管理驱动设置统一

“exec-opts”: [“native.cgroupdriver=systemd”]

root@master1:~# vim /etc/docker/daemon.json

{

 "registry-mirrors": ["https://9y0z4v79.mirror.aliyuncs.com"],

 "exec-opts": ["native.cgroupdriver=systemd"]

}

重刷配置文件生效：

systemctl daemon-reload

重启docker

systemctl restart docker

重启集群

docker exec -it kubeasz ezctl setup default all