kubeadm部署高可用案例中api-server连接etcd问题

[76439984jc]

我发现如果以kubeadm这样的方式来做高可用的，他每个master上的api-server的pod里面连接etcd的参数都会是 --etcd-servers=https://127.0.0.1:2379而不是etcd上的节点ip地址，这样会不会有问题的，因为当api-server没问题但是本地etc有问题时候那这个master始终会访问不了etcd了。但如果强行把这个改掉或者单独将这个部署成外部etcd的话，那证书不会很难搞呢。

[chhsiao1981]

This is design choice～

If there is something wrong with the local etcd, it's highly possible that there is some malfunctioning in the machine, and is not recommended to serve as part of the master-node.

[76439984jc]

This is design choice～

If there is something wrong with the local etcd, it's highly possible that there is some malfunctioning in the machine, and is not recommended to serve as part of the master-node.

但有没有人是部署好以后进行修改的呢，还有现在生产环境上用的二进制部署的比较多还是kubeadm呢？

[chhsiao1981]

You can check the "external-etcd setting" for kubeadm.
It may fit your requirement:

https://kubernetes.io/docs/setup/production-environment/tools/kubeadm/high-availability/#external-etcd-nodes

[76439984jc]

You can check the "external-etcd setting" for kubeadm.
It may fit your requirement:

https://kubernetes.io/docs/setup/production-environment/tools/kubeadm/high-availability/#external-etcd-nodes

好的。非常感谢，我也正在看这个配置考虑是否在外部部署etcd