【k8s】单master到多master平滑升级

小洋-

已于 2023-04-21 14:27:21 修改

阅读量1.6k

点赞数 2

文章标签： kubernetes docker 容器

于 2022-08-30 19:04:27 首次发布

本文链接：https://blog.csdn.net/weixin_45066823/article/details/126609892

版权

升级原因

由于单master节点用于线上环境发生故障时会导致集群整个不可用，生产环境中有很大风险所以要把集群升级为多master且高可用，其中任何一个节点宕机都不影响集群正常运行。

本机环境

mater  control-plane,master v1.21.0
node1  node                 v1.21.0
node2  node                 v1.21.0

计划将node1和node2升级为master。为了api-server高可用，在另两台机器上部署了haproxy用来代理api-server。后文用inspot:6443表示新的api-server 地址。

备份k8s环境

以防万一备份k8s配置

cp -a /etc/kubernetes  /etc/kubernetes.bak

申请新的api-server证书

将 kube-system 中的 kubeadm-config 配置导出，并修改配置

kubectl -n kube-system get configmap kubeadm-config -o jsonpath='{.data.ClusterConfiguration}' > kubeadm.yaml

旧配置

apiServer:
  extraArgs:
    authorization-mode: Node,RBAC
  timeoutForControlPlane: 4m0s
apiVersion: kubeadm.k8s.io/v1beta2
certificatesDir: /etc/kubernetes/pki
clusterName: kubernetes
controlPlaneEndpoint: master:6443
controllerManager: {}
dns:
  type: CoreDNS
etcd:
  local:
    dataDir: /var/lib/etcd
imageRepository: k8s.gcr.io
kind: ClusterConfiguration
kubernetesVersion: v1.21.0
networking:
  dnsDomain: cluster.local
  podSubnet: 192.168.0.0/16
  serviceSubnet: 10.96.0.0/12
scheduler: {}

新的配置

apiServer:
  certSANs:
  - inspot
  - master
  - node1
  - node2
  - 10.196.96.11
  - 10.196.96.12
  - 10.196.96.17
  extraArgs:
    authorization-mode: Node,RBAC
  timeoutForControlPlane: 4m0s
apiVersion: kubeadm.k8s.io/v1beta2
certificatesDir: /etc/kubernetes/pki
clusterName: kubernetes
controlPlaneEndpoint: inspot:6443
controllerManager: {}
dns:
  type: CoreDNS
etcd:
  local:
    dataDir: /var/lib/etcd
imageRepository: k8s.gcr.io
kind: ClusterConfiguration
kubernetesVersion: v1.21.0
networking:
  dnsDomain: cluster.local
  podSubnet: 192.168.0.0/16
  serviceSubnet: 10.96.0.0/12
scheduler: {}

生成apiserver证书

移除并备份旧证书

mv /etc/kubernetes/pki/apiserver.{crt,key}  /备份目录

生成新的证书

kubeadm init phase certs apiserver --config kubeadm.yaml

验证证书内容

openssl x509 -in /etc/kubernetes/pki/apiserver.crt -text

发现新的证书已经更新的dns部分

Certificate:
    Data:
        Version: 3 (0x2)
        Serial Number:
            20:be:05:25:b1:76:88:59:48:e1:bb:60:36:06:8e:8e:33:d2:e4:e2
        Signature Algorithm: sha256WithRSAEncryption
        Issuer: CN = kubernetes
        Validity
            Not Before: Aug 30 02:16:10 2022 GMT
            Not After : Aug 27 02:16:10 2032 GMT
        Subject: CN = kube-apiserver
        Subject Public Key Info:
            Public Key Algorithm: rsaEncryption
                RSA Public-Key: (2048 bit)
                Modulus:
                    00:a9:64:9e:8d:f7:93:b0:48:aa:b0:21:d9:db:ea:
                    17:19:ae:2e:f0:40:b0:25:d8:c2:6c:26:7a:2e:48:
                    f7:93:51:96:d4:1c:d8:0a:b2:57:bc:49:f5:4a:73:
                    d3:f6:6c:f2:b0:98:14:05:91:47:87:11:a4:0e:de:
                    0c:6e:4e:6c:59:37:2f:50:0c:23:53:0b:18:6e:66:
                    c0:c7:64:88:c4:2d:93:ce:13:3b:c4:0f:89:68:2a:
                    48:09:ac:91:46:5b:c9:f8:f1:48:10:3b:10:fd:70:
                    20:d0:b4:a2:62:c1:0b:b0:91:2c:10:83:06:e6:df:
                    fc:54:10:da:32:d3:db:79:e7:3e:55:c2:41:92:33:
                    d1:3e:14:9d:d7:c4:f1:02:40:88:e7:2d:97:b9:26:
                    20:a0:9e:65:05:d6:51:90:22:98:3a:19:ce:56:cf:
                    68:d0:d5:98:33:e0:af:5c:df:20:f8:b6:83:60:2b:
                    fb:af:2b:86:33:c4:0b:b6:a6:1e:d9:ce:14:4e:54:
                    48:12:10:2d:51:0f:db:bb:9b:e0:e0:4d:a9:17:b2:
                    8b:a3:44:74:15:75:76:92:4e:95:34:9c:b7:2b:91:
                    6c:8f:80:d9:c4:5c:1e:24:39:59:40:fb:5b:9e:2e:
                    bb:fc:87:e3:47:cd:1c:05:22:d3:b0:62:9c:1e:a1:
                    1d:f7
                Exponent: 65537 (0x10001)
        X509v3 extensions:
            X509v3 Key Usage: critical
                Digital Signature, Key Encipherment
            X509v3 Extended Key Usage:
                TLS Web Server Authentication
            X509v3 Subject Alternative Name:
                DNS:inspot, DNS:kubernetes, DNS:kubernetes.default, DNS:kubernetes.default.svc, DNS:kubernetes.default.svc.cluster.local, DNS:master, DNS:node1, DNS:node2, IP Address:10.96.0.1, IP Address:10.196.96.11, IP Address:10.196.96.12, IP Address:10.196.96.17
    Signature Algorithm: sha256WithRSAEncryption
         7b:11:fb:0e:3d:9c:e7:79:d9:e9:1a:af:06:62:d9:39:ee:fe:
         d0:26:bb:5d:7d:51:b5:0c:f1:f0:e9:db:d6:37:46:c4:39:cc:
         38:1e:20:99:60:d1:21:f8:5c:9d:73:81:ec:d6:cf:c1:24:25:
         30:69:42:6c:80:04:8a:85:e7:f3:6e:de:01:61:f4:d4:e3:4f:
         a5:d1:28:5f:43:e8:2d:15:f9:e2:29:51:15:87:d3:66:99:84:
         3b:31:6b:62:ee:98:71:b1:25:bf:fc:c8:e2:b5:d5:0b:ee:a1:
         a2:59:b6:f8:fb:d8:94:7c:6a:f8:3e:f3:bc:2d:e1:01:68:58:
         34:06:5e:ab:88:07:44:1f:bb:a1:83:2d:bb:ea:58:c3:ee:ed:
         14:9f:1b:d3:a7:56:bd:08:a4:a5:74:2a:ea:9b:45:d1:e4:fd:
         5f:40:00:30:be:dc:5a:72:e6:c5:4b:e3:6f:3b:a3:84:d8:86:
         84:2f:6f:a4:d4:45:a7:15:14:ff:18:8e:ff:4b:42:4a:eb:b0:
         18:fb:32:95:ae:d2:24:dc:c2:5c:35:2e:68:df:4b:51:46:ae:
         8d:8e:9c:44:d0:09:a6:eb:3a:76:30:3f:f5:8c:f3:4a:59:1c:
         cc:e7:d9:ac:f2:36:07:ca:96:51:0e:0c:3d:5b:da:f4:ac:36:
         50:2f:68:a7
-----BEGIN CERTIFICATE-----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-----END CERTIFICATE-----

新增的内容回填 kubeadm-config

kubectl edit cm kubeadm-config -n kube-system
将这部分再写一遍
  certSANs:
  - inspot
  - master
  - node1
  - node2
  - 10.196.96.11
  - 10.196.96.12
  - 10.196.96.17

替换api-server地址

# 模版
sed -i "s/oldapiservername/newapiservername/g" `grep oldapiservername -rl /etc/kubernetes`
# 示例
sed -i "s/10.196.96.11:6443/inspot:6443/g" `grep "10.196.96.11:6443" -rl /etc/kubernetes`

修改默认配置文件中的api-server地址

vi ~/.kube/config

修改kubeadm-config中api-server地址

kubectl edit cm kubeadm-config -n kube-system

controlPlaneEndpoint属性inspot:6443

重启api-server等相关容器

docker ps |grep -E 'k8s_kube-apiserver|k8s_kube-controller-manager|k8s_kube-scheduler|k8s_etcd_etcd' | awk -F ' ' '{print $1}' |xargs docker restart

更新cluster-info

kubectl -n kube-public edit cm cluster-info

server: https://inspot:6443

验证 cluster-info，显示此输出即可

root@master:~# kubectl cluster-info
Kubernetes control plane is running at https://inspot:6443
CoreDNS is running at https://inspot:6443/api/v1/namespaces/kube-system/services/kube-dns:dns/proxy

To further debug and diagnose cluster problems, use 'kubectl cluster-info dump'.

上传证书到集群，此处输出的key在后面有用

kubeadm init phase upload-certs --upload-certs --config kubeadm.yaml



I0612 02:21:45.938155  103642 version.go:252] remote version is much newer: v1.24.1; falling back to: stable-1.19
W0612 02:21:46.570837  103642 configset.go:348] WARNING: kubeadm cannot validate component configs for API groups [kubelet.config.k8s.io kubeproxy.config.k8s.io]
[upload-certs] Storing the certificates in Secret "kubeadm-certs" in the "kube-system" Namespace
[upload-certs] Using certificate key:
# 此处生成的key在后续master加入集群时可用
1b594f631654ac0f221dd0cf593605a30b8f41eee480562d2a3103ab667cfcc7

下面的命令会生成一个token和命令后面会用到

kubeadm token create --print-join-command --config kubeadm.yaml

根据上面操作的信息屏凑一个加入主节点的命令，在新加入的2个主节点执行

kubeadm join inspot:6443 \
--token f27w7m.adelvl3waw9kqdhp \  # 这个token可以通过kubeadm token create创建
--discovery-token-ca-cert-hash <第二步生成的token> \
--control-plane --certificate-key <第一步生成的key>

因为其他两个node节点在使用中，所以需要先设置节点不可调度，再驱逐删除那个节点

kubectl cordon node1
# --delete-emptydir-data会删除statefulset（有状态应用）
kubectl drain node1 --ignore-daemonsets --force --delete-emptydir-data
kubectl delete node node1

因为被删除的节点不是很“干净”，所以还需要把/etc/kubernets下的内容移除

kubeadm reset
mv /etc/kubernets/* 备份目录
mv /var/lib/etcd/* 备份目录

随后执行上面拼凑的kubeadm join即可！

小洋-

关注

2
点赞
踩
11

收藏

觉得还不错? 一键收藏
0
评论
【k8s】单master到多master平滑升级

由于单master节点用于线上环境发生故障时会导致集群整个不可用，生产环境中有很大风险所以要把集群升级为多master模式，其中任何一个节点宕机都不影响集群正常运行。将 kube-system 中的 kubeadm-config 配置导出，并修改配置。因为被删除的节点不是很“干净”，所以还需要把/etc/kubernets下的内容移除。根据上面操作的信息屏凑一个加入主节点的命令，在新加入的2个主节点执行。修改kubeadm-config中api-server地址。发现新的证书已经更新的。.........
复制链接

扫一扫