k8s master节点迁移

环境概况

master节点涉及组建

kube-apiserver
kube-scheduler
kube-controller-manager
flanneld

node节点涉及组建

kubelet
kube-proxy
flanneld

etcd集群

etcd集群迁移

master节点迁移

部署完新的master节点后,修改旧master节点配置

改旧的apiserver 10.10.64.144 组件配置,加上选主等配置
首先disable !!! 
systemctl disable kube-apiserver
systemctl disable kube-scheduler
systemctl disable kube-controller
修改为
/etc/sysconfig/kube-apiserver
KUBE_APISERVER_OPTS='--apiserver-count=2'
修改为
/etc/sysconfig/kube-scheduler
KUBE_SCHEDULER_OPTS='--leader-elect=true'
修改为
/etc/sysconfig/kube-controller
KUBE_CONTROLLER_OPTS='--cloud-provider= --leader-elect=true'
修改kube-apiserver的etcd地址
ETCD_SERVERS='--etcd-servers=新的etcd地址'
systemctl restart kube-apiserver
systemctl restart kube-scheduler
systemctl restart kube-controller

修改新的apiserver组件配置,加上选主等配置

修改为
/etc/sysconfig/kube-apiserver
KUBE_APISERVER_OPTS='--apiserver-count=2'
修改为
/etc/sysconfig/kube-scheduler
KUBE_SCHEDULER_OPTS='--leader-elect=true'
修改为
/etc/sysconfig/kube-controller
KUBE_CONTROLLER_OPTS='--cloud-provider= --leader-elect=true'
修改kube-apiserver的etcd地址
ETCD_SERVERS='--etcd-servers=新的etcd地址'
systemctl restart kube-apiserver
systemctl restart kube-scheduler
systemctl restart kube-controller

修改所有节点flannel的etcd地址


禁用flanneld
ansible -i inventory/appserver/online_node.yml online_node -m shell -a "systemctl disable flanneld"
修改flanneld地址
ansible -i inventory/appserver/online_node.yml online_node -m shell -a "sed -i 's/old_ip/new_ip/g' /etc/sysconfig/flanneld"
ansible -i inventory/appserver/online_node.yml online_node -m shell -a "sed -i 's/old_ip/new_ip/g' /etc/sysconfig/flanneld"
重启flanneld
ansible -i inventory/appserver/online_node.yml online_node -m shell -a "systemctl restart flanneld"
开机启动flanneld
ansible -i inventory/appserver/online_node.yml online_node -m shell -a "systemctl enable flanneld"
如果回滚,调换下ip位置。按同样方式执行。

同理修改kubelet,kube-proxy的配置文件

观察一段时间,测试功能,一切没问题的话,停止旧的apiserver。

去掉新apiserver 选主配置

切换观察如果没问题后,关机迁移旧的 master节点后,去掉新的apiserver选主等配置重启。


本博客所有文章除特别声明外,均采用 CC BY-SA 4.0 协议 ,转载请注明出处!