【k8s部署】6. 部署 worker 节点

时间 2020-07-07

标签 k8s部署部署 worker 节点繁體版

原文原文链接

若是没有特殊指明，全部操做均在 zhaoyixin-k8s-01 节点上执行。html

kubernetes worker 节点运行以下组件：node

containerd
kubelet
kube-proxy
calico
kube-nginx

0.安装依赖包

source /opt/k8s/bin/environment.sh
for node_ip in ${NODE_IPS[@]}
  do
    echo ">>> ${node_ip}"
    ssh root@${node_ip} "yum install -y epel-release" &
    ssh root@${node_ip} "yum install -y chrony conntrack ipvsadm ipset jq iptables curl sysstat libseccomp wget socat git" &
  done

1.apiserver 高可用

本节使用 nginx 4 层透明代理功能实现 Kubernetes worker 节点组件高可用访问 kube-apiserver 集群。linux

基于 nginx 代理的 kube-apiserver 高可用方案

控制节点的 kube-controller-manager、kube-scheduler 是多实例部署且链接本机的 kube-apiserver，因此只要有一个实例正常，就能够保证高可用；
集群内的 Pod 使用 K8S 服务域名 kubernetes 访问 kube-apiserver， kube-dns 会自动解析出多个 kube-apiserver 节点的 IP，因此也是高可用的；
在每一个节点起一个 nginx 进程，后端对接多个 apiserver 实例，nginx 对它们作健康检查和负载均衡；
kubelet、kube-proxy 经过本地的 nginx（监听 127.0.0.1）访问 kube-apiserver，从而实现 kube-apiserver 的高可用。

下载和编译 nginx

下载源码：nginx

cd /opt/k8s/work
wget http://nginx.org/download/nginx-1.15.3.tar.gz
tar -xzvf nginx-1.15.3.tar.gz

配置编译参数：git

cd /opt/k8s/work/nginx-1.15.3
mkdir nginx-prefix
yum install -y gcc make
./configure --with-stream --without-http --prefix=$(pwd)/nginx-prefix --without-http_uwsgi_module --without-http_scgi_module --without-http_fastcgi_module

--with-stream：开启 4 层透明转发(TCP Proxy)功能；
--without-xxx：关闭全部其余功能，这样生成的动态连接二进制程序依赖最小；

输出：github

Configuration summary
  + PCRE library is not used
  + OpenSSL library is not used
  + zlib library is not used

  nginx path prefix: "/opt/k8s/work/nginx-1.15.3/nginx-prefix"
  nginx binary file: "/opt/k8s/work/nginx-1.15.3/nginx-prefix/sbin/nginx"
  nginx modules path: "/opt/k8s/work/nginx-1.15.3/nginx-prefix/modules"
  nginx configuration prefix: "/opt/k8s/work/nginx-1.15.3/nginx-prefix/conf"
  nginx configuration file: "/opt/k8s/work/nginx-1.15.3/nginx-prefix/conf/nginx.conf"
  nginx pid file: "/opt/k8s/work/nginx-1.15.3/nginx-prefix/logs/nginx.pid"
  nginx error log file: "/opt/k8s/work/nginx-1.15.3/nginx-prefix/logs/error.log"
  nginx http access log file: "/opt/k8s/work/nginx-1.15.3/nginx-prefix/logs/access.log"
  nginx http client request body temporary files: "client_body_temp"
  nginx http proxy temporary files: "proxy_temp"

编译和安装：web

cd /opt/k8s/work/nginx-1.15.3
make && make install

验证编译安装的 nginx正则表达式

cd /opt/k8s/work/nginx-1.15.3
./nginx-prefix/sbin/nginx -v
nginx version: nginx/1.15.3

安装和部署 nginx

建立目录结构：docker

cd /opt/k8s/work
source /opt/k8s/bin/environment.sh
for node_ip in ${NODE_IPS[@]}
  do
    echo ">>> ${node_ip}"
    ssh root@${node_ip} "mkdir -p /opt/k8s/kube-nginx/{conf,logs,sbin}"
  done

拷贝二进制程序，并重命名为 kube-nginx。json

cd /opt/k8s/work
source /opt/k8s/bin/environment.sh
for node_ip in ${NODE_IPS[@]}
  do
    echo ">>> ${node_ip}"
    ssh root@${node_ip} "mkdir -p /opt/k8s/kube-nginx/{conf,logs,sbin}"
    scp /opt/k8s/work/nginx-1.15.3/nginx-prefix/sbin/nginx  root@${node_ip}:/opt/k8s/kube-nginx/sbin/kube-nginx
    ssh root@${node_ip} "chmod a+x /opt/k8s/kube-nginx/sbin/*"
  done

配置 nginx，开启 4 层透明转发功能：

cd /opt/k8s/work
cat > kube-nginx.conf << \EOF
worker_processes 1;

events {
    worker_connections  1024;
}

stream {
    upstream backend {
        hash $remote_addr consistent;
        server 192.168.16.8:6443        max_fails=3 fail_timeout=30s;
        server 192.168.16.10:6443        max_fails=3 fail_timeout=30s;
        server 192.168.16.6:6443        max_fails=3 fail_timeout=30s;
    }

    server {
        listen 127.0.0.1:8443;
        proxy_connect_timeout 1s;
        proxy_pass backend;
    }
}
EOF

upstream backend 中的 server 列表为集群中各 kube-apiserver 的节点 IP，须要根据实际状况修改；

分发配置文件：

cd /opt/k8s/work
source /opt/k8s/bin/environment.sh
for node_ip in ${NODE_IPS[@]}
  do
    echo ">>> ${node_ip}"
    scp kube-nginx.conf  root@${node_ip}:/opt/k8s/kube-nginx/conf/kube-nginx.conf
  done

配置 systemd unit 文件，启动服务

配置 kube-nginx systemd unit 文件：

cd /opt/k8s/work
cat > kube-nginx.service <<EOF
[Unit]
Description=kube-apiserver nginx proxy
After=network.target
After=network-online.target
Wants=network-online.target

[Service]
Type=forking
ExecStartPre=/opt/k8s/kube-nginx/sbin/kube-nginx -c /opt/k8s/kube-nginx/conf/kube-nginx.conf -p /opt/k8s/kube-nginx -t
ExecStart=/opt/k8s/kube-nginx/sbin/kube-nginx -c /opt/k8s/kube-nginx/conf/kube-nginx.conf -p /opt/k8s/kube-nginx
ExecReload=/opt/k8s/kube-nginx/sbin/kube-nginx -c /opt/k8s/kube-nginx/conf/kube-nginx.conf -p /opt/k8s/kube-nginx -s reload
PrivateTmp=true
Restart=always
RestartSec=5
StartLimitInterval=0
LimitNOFILE=65536

[Install]
WantedBy=multi-user.target
EOF

分发 systemd unit 文件：

cd /opt/k8s/work
source /opt/k8s/bin/environment.sh
for node_ip in ${NODE_IPS[@]}
  do
    echo ">>> ${node_ip}"
    scp kube-nginx.service  root@${node_ip}:/etc/systemd/system/
  done

启动 kube-nginx 服务：

cd /opt/k8s/work
source /opt/k8s/bin/environment.sh
for node_ip in ${NODE_IPS[@]}
  do
    echo ">>> ${node_ip}"
    ssh root@${node_ip} "systemctl daemon-reload && systemctl enable kube-nginx && systemctl restart kube-nginx"
  done

检查 kube-nginx 服务运行状态

cd /opt/k8s/work
source /opt/k8s/bin/environment.sh
for node_ip in ${NODE_IPS[@]}
  do
    echo ">>> ${node_ip}"
    ssh root@${node_ip} "systemctl status kube-nginx |grep 'Active:'"
  done

确保状态为 active (running)，不然经过journalctl -u kube-nginx查看日志，确认缘由。

2.部署 containerd 组件

containerd 实现了 kubernetes 的 Container Runtime Interface (CRI) 接口，提供容器运行时核心功能，如镜像管理、容器管理等，相比 dockerd 更加简单、健壮和可移植。

下载和分发二进制文件

下载二进制文件：

cd /opt/k8s/work
wget https://github.com/kubernetes-sigs/cri-tools/releases/download/v1.17.0/crictl-v1.17.0-linux-amd64.tar.gz \
  https://github.com/opencontainers/runc/releases/download/v1.0.0-rc10/runc.amd64 \
  https://github.com/containernetworking/plugins/releases/download/v0.8.5/cni-plugins-linux-amd64-v0.8.5.tgz \
  https://github.com/containerd/containerd/releases/download/v1.3.3/containerd-1.3.3.linux-amd64.tar.gz

解压：

cd /opt/k8s/work
mkdir containerd
tar -xvf containerd-1.3.3.linux-amd64.tar.gz -C containerd
tar -xvf crictl-v1.17.0-linux-amd64.tar.gz

mkdir cni-plugins
sudo tar -xvf cni-plugins-linux-amd64-v0.8.5.tgz -C cni-plugins

sudo mv runc.amd64 runc

分发二进制文件到全部 worker 节点：

cd /opt/k8s/work
source /opt/k8s/bin/environment.sh
for node_ip in ${NODE_IPS[@]}
  do
    echo ">>> ${node_ip}"
    scp containerd/bin/*  crictl  cni-plugins/*  runc  root@${node_ip}:/opt/k8s/bin
    ssh root@${node_ip} "chmod a+x /opt/k8s/bin/* && mkdir -p /etc/cni/net.d"
  done

建立和分发 containerd 配置文件

cd /opt/k8s/work
source /opt/k8s/bin/environment.sh
cat << EOF | sudo tee containerd-config.toml
version = 2
root = "${CONTAINERD_DIR}/root"
state = "${CONTAINERD_DIR}/state"

[plugins]
  [plugins."io.containerd.grpc.v1.cri"]
    sandbox_image = "registry.cn-beijing.aliyuncs.com/images_k8s/pause-amd64:3.1"
    [plugins."io.containerd.grpc.v1.cri".cni]
      bin_dir = "/opt/k8s/bin"
      conf_dir = "/etc/cni/net.d"
  [plugins."io.containerd.runtime.v1.linux"]
    shim = "containerd-shim"
    runtime = "runc"
    runtime_root = ""
    no_shim = false
    shim_debug = false
EOF

cd /opt/k8s/work
source /opt/k8s/bin/environment.sh
for node_ip in ${NODE_IPS[@]}
  do
    echo ">>> ${node_ip}"
    ssh root@${node_ip} "mkdir -p /etc/containerd/ ${CONTAINERD_DIR}/{root,state}"
    scp containerd-config.toml root@${node_ip}:/etc/containerd/config.toml
  done

建立 containerd systemd unit 文件

cd /opt/k8s/work
cat <<EOF | sudo tee containerd.service
[Unit]
Description=containerd container runtime
Documentation=https://containerd.io
After=network.target

[Service]
Environment="PATH=/opt/k8s/bin:/bin:/sbin:/usr/bin:/usr/sbin"
ExecStartPre=/sbin/modprobe overlay
ExecStart=/opt/k8s/bin/containerd
Restart=always
RestartSec=5
Delegate=yes
KillMode=process
OOMScoreAdjust=-999
LimitNOFILE=1048576
LimitNPROC=infinity
LimitCORE=infinity

[Install]
WantedBy=multi-user.target
EOF

分发 systemd unit 文件，启动 containerd 服务

cd /opt/k8s/work
source /opt/k8s/bin/environment.sh
for node_ip in ${NODE_IPS[@]}
  do
    echo ">>> ${node_ip}"
    scp containerd.service root@${node_ip}:/etc/systemd/system
    ssh root@${node_ip} "systemctl enable containerd && systemctl restart containerd"
  done

建立和分发 crictl 配置文件

crictl 是兼容 CRI 容器运行时的命令行工具，提供相似于 docker 命令的功能。具体参考官方文档。

cd /opt/k8s/work
cat << EOF | sudo tee crictl.yaml
runtime-endpoint: unix:///run/containerd/containerd.sock
image-endpoint: unix:///run/containerd/containerd.sock
timeout: 10
debug: false
EOF

分发到全部 worker 节点：

cd /opt/k8s/work
source /opt/k8s/bin/environment.sh
for node_ip in ${NODE_IPS[@]}
  do
    echo ">>> ${node_ip}"
    scp crictl.yaml root@${node_ip}:/etc/crictl.yaml
  done

3.部署 kubelet 组件

kubelet 运行在每一个 worker 节点上，接收 kube-apiserver 发送的请求，管理 Pod 容器，执行交互式命令，如 exec、run、logs 等。

kubelet 启动时自动向 kube-apiserver 注册节点信息，内置的 cadvisor 统计和监控节点的资源使用状况。

为确保安全，部署时关闭了 kubelet 的非安全 http 端口，对请求进行认证和受权，拒绝未受权的访问(如 apiserver、heapster 的请求)。

下载和分发 kubelet 二进制文件

参考【k8s部署】5. 部署 master 节点中的《下载二进制文件》一节。

建立 kubelet bootstrap kubeconfig 文件

cd /opt/k8s/work
source /opt/k8s/bin/environment.sh
for node_name in ${NODE_NAMES[@]}
  do
    echo ">>> ${node_name}"

    # 建立 token
    export BOOTSTRAP_TOKEN=$(kubeadm token create \
      --description kubelet-bootstrap-token \
      --groups system:bootstrappers:${node_name} \
      --kubeconfig ~/.kube/config)

    # 设置集群参数
    kubectl config set-cluster kubernetes \
      --certificate-authority=/etc/kubernetes/cert/ca.pem \
      --embed-certs=true \
      --server=${KUBE_APISERVER} \
      --kubeconfig=kubelet-bootstrap-${node_name}.kubeconfig

    # 设置客户端认证参数
    kubectl config set-credentials kubelet-bootstrap \
      --token=${BOOTSTRAP_TOKEN} \
      --kubeconfig=kubelet-bootstrap-${node_name}.kubeconfig

    # 设置上下文参数
    kubectl config set-context default \
      --cluster=kubernetes \
      --user=kubelet-bootstrap \
      --kubeconfig=kubelet-bootstrap-${node_name}.kubeconfig

    # 设置默认上下文
    kubectl config use-context default --kubeconfig=kubelet-bootstrap-${node_name}.kubeconfig
  done

向 kubeconfig 写入的是 token，bootstrap 结束后 kube-controller-manager 为 kubelet 建立 client 和 server 证书；

查看 kubeadm 为各节点建立的 token：

$ kubeadm token list --kubeconfig ~/.kube/config
TOKEN                     TTL       EXPIRES                     USAGES                   DESCRIPTION               EXTRA GROUPS
3uk8cy.1yeeawz00uxr2r01   23h       2020-05-31T15:05:58+08:00   authentication,signing   kubelet-bootstrap-token   system:bootstrappers:zhaoyixin-k8s-03
udg7tq.qh9dksbq0u0jxjat   23h       2020-05-31T15:05:55+08:00   authentication,signing   kubelet-bootstrap-token   system:bootstrappers:zhaoyixin-k8s-01
vl120m.v8a8hdecwkpo4cyn   23h       2020-05-31T15:05:57+08:00   authentication,signing   kubelet-bootstrap-token   system:bootstrappers:zhaoyixin-k8s-02

token 有效期为 1 天，超期后将不能再被用来 boostrap kubelet，且会被 kube-controller-manager 的 tokencleaner 清理；
kube-apiserver 接收 kubelet 的 bootstrap token 后，将请求的 user 设置为 system:bootstrap:<Token ID>，group 设置为 system:bootstrappers，后续将为这个 group 设置 ClusterRoleBinding；

分发 bootstrap kubeconfig 文件到全部 worker 节点

cd /opt/k8s/work
source /opt/k8s/bin/environment.sh
for node_name in ${NODE_NAMES[@]}
  do
    echo ">>> ${node_name}"
    scp kubelet-bootstrap-${node_name}.kubeconfig root@${node_name}:/etc/kubernetes/kubelet-bootstrap.kubeconfig
  done

建立和分发 kubelet 参数配置文件

建立 kubelet 参数配置文件模板（可配置项参考代码中注释）：

cd /opt/k8s/work
source /opt/k8s/bin/environment.sh
cat > kubelet-config.yaml.template <<EOF
kind: KubeletConfiguration
apiVersion: kubelet.config.k8s.io/v1beta1
address: "##NODE_IP##"
staticPodPath: ""
syncFrequency: 1m
fileCheckFrequency: 20s
httpCheckFrequency: 20s
staticPodURL: ""
port: 10250
readOnlyPort: 0
rotateCertificates: true
serverTLSBootstrap: true
authentication:
  anonymous:
    enabled: false
  webhook:
    enabled: true
  x509:
    clientCAFile: "/etc/kubernetes/cert/ca.pem"
authorization:
  mode: Webhook
registryPullQPS: 0
registryBurst: 20
eventRecordQPS: 0
eventBurst: 20
enableDebuggingHandlers: true
enableContentionProfiling: true
healthzPort: 10248
healthzBindAddress: "##NODE_IP##"
clusterDomain: "${CLUSTER_DNS_DOMAIN}"
clusterDNS:
  - "${CLUSTER_DNS_SVC_IP}"
nodeStatusUpdateFrequency: 10s
nodeStatusReportFrequency: 1m
imageMinimumGCAge: 2m
imageGCHighThresholdPercent: 85
imageGCLowThresholdPercent: 80
volumeStatsAggPeriod: 1m
kubeletCgroups: ""
systemCgroups: ""
cgroupRoot: ""
cgroupsPerQOS: true
cgroupDriver: cgroupfs
runtimeRequestTimeout: 10m
hairpinMode: promiscuous-bridge
maxPods: 220
podCIDR: "${CLUSTER_CIDR}"
podPidsLimit: -1
resolvConf: /etc/resolv.conf
maxOpenFiles: 1000000
kubeAPIQPS: 1000
kubeAPIBurst: 2000
serializeImagePulls: false
evictionHard:
  memory.available:  "100Mi"
  nodefs.available:  "10%"
  nodefs.inodesFree: "5%"
  imagefs.available: "15%"
evictionSoft: {}
enableControllerAttachDetach: true
failSwapOn: true
containerLogMaxSize: 20Mi
containerLogMaxFiles: 10
systemReserved: {}
kubeReserved: {}
systemReservedCgroup: ""
kubeReservedCgroup: ""
enforceNodeAllocatable: ["pods"]
EOF

address：kubelet 安全端口（https，10250）监听的地址，不能为 127.0.0.1，不然 kube-apiserver、heapster 等不能调用 kubelet 的 API；
readOnlyPort=0：关闭只读端口(默认 10255)，等效为未指定；
authentication.anonymous.enabled：设置为 false，不容许匿名访问 10250 端口；
authentication.x509.clientCAFile：指定签名客户端证书的 CA 证书，开启 HTTP 证书认证；
authentication.webhook.enabled=true：开启 HTTPs bearer token 认证；
对于未经过 x509 证书和 webhook 认证的请求(kube-apiserver 或其余客户端)，将被拒绝，提示 Unauthorized；
authroization.mode=Webhook：kubelet 使用 SubjectAccessReview API 查询 kube-apiserver 某 user、group 是否具备操做资源的权限(RBAC)；
featureGates.RotateKubeletClientCertificate、featureGates.RotateKubeletServerCertificate：自动 rotate 证书，证书的有效期取决于 kube-controller-manager 的 --experimental-cluster-signing-duration 参数；
须要 root 帐户运行；

为各节点建立和分发 kubelet 配置文件：

cd /opt/k8s/work
source /opt/k8s/bin/environment.sh
for node_ip in ${NODE_IPS[@]}
  do 
    echo ">>> ${node_ip}"
    sed -e "s/##NODE_IP##/${node_ip}/" kubelet-config.yaml.template > kubelet-config-${node_ip}.yaml.template
    scp kubelet-config-${node_ip}.yaml.template root@${node_ip}:/etc/kubernetes/kubelet-config.yaml
  done

建立和分发 kubelet systemd unit 文件

建立 kubelet systemd unit 文件模板：

cd /opt/k8s/work
source /opt/k8s/bin/environment.sh
cat > kubelet.service.template <<EOF
[Unit]
Description=Kubernetes Kubelet
Documentation=https://github.com/GoogleCloudPlatform/kubernetes
After=containerd.service
Requires=containerd.service

[Service]
WorkingDirectory=${K8S_DIR}/kubelet
ExecStart=/opt/k8s/bin/kubelet \\
  --bootstrap-kubeconfig=/etc/kubernetes/kubelet-bootstrap.kubeconfig \\
  --cert-dir=/etc/kubernetes/cert \\
  --network-plugin=cni \\
  --cni-conf-dir=/etc/cni/net.d \\
  --container-runtime=remote \\
  --container-runtime-endpoint=unix:///var/run/containerd/containerd.sock \\
  --root-dir=${K8S_DIR}/kubelet \\
  --kubeconfig=/etc/kubernetes/kubelet.kubeconfig \\
  --config=/etc/kubernetes/kubelet-config.yaml \\
  --hostname-override=##NODE_NAME## \\
  --image-pull-progress-deadline=15m \\
  --volume-plugin-dir=${K8S_DIR}/kubelet/kubelet-plugins/volume/exec/ \\
  --logtostderr=true \\
  --v=2
Restart=always
RestartSec=5
StartLimitInterval=0

[Install]
WantedBy=multi-user.target
EOF

若是设置了 --hostname-override 选项，则 kube-proxy 也须要设置该选项，不然会出现找不到 Node 的状况；
--bootstrap-kubeconfig：指向 bootstrap kubeconfig 文件，kubelet 使用该文件中的用户名和 token 向 kube-apiserver 发送 TLS Bootstrapping 请求；
K8S approve kubelet 的 csr 请求后，在 --cert-dir 目录建立证书和私钥文件，而后写入 --kubeconfig 文件；
--pod-infra-container-image 不使用 redhat 的 pod-infrastructure:latest 镜像，它不能回收容器的僵尸；

为各节点建立和分发 kubelet systemd unit 文件：

cd /opt/k8s/work
source /opt/k8s/bin/environment.sh
for node_name in ${NODE_NAMES[@]}
  do 
    echo ">>> ${node_name}"
    sed -e "s/##NODE_NAME##/${node_name}/" kubelet.service.template > kubelet-${node_name}.service
    scp kubelet-${node_name}.service root@${node_name}:/etc/systemd/system/kubelet.service
  done

授予 kube-apiserver 访问 kubelet API 的权限

在执行 kubectl exec、run、logs 等命令时，apiserver 会将请求转发到 kubelet 的 https 端口。这里定义 RBAC 规则，受权 apiserver 使用的证书（kubernetes.pem）用户名（CN：kuberntes-master）访问 kubelet API 的权限：

kubectl create clusterrolebinding kube-apiserver:kubelet-apis --clusterrole=system:kubelet-api-admin --user kubernetes-master

Bootstrap Token Auth 和授予权限

kubelet 启动时查找 --kubeletconfig 参数对应的文件是否存在，若是不存在则使用 --bootstrap-kubeconfig 指定的 kubeconfig 文件向 kube-apiserver 发送证书签名请求 (CSR)。

kube-apiserver 收到 CSR 请求后，对其中的 Token 进行认证，认证经过后将请求的 user 设置为 system:bootstrap:<Token ID>，group 设置为 system:bootstrappers，这一过程称为 Bootstrap Token Auth。

默认状况下，这个 user 和 group 没有建立 CSR 的权限，kubelet 启动失败。

解决办法是：建立一个 clusterrolebinding，将 group system:bootstrappers 和 clusterrole system:node-bootstrapper 绑定：

kubectl create clusterrolebinding kubelet-bootstrap --clusterrole=system:node-bootstrapper --group=system:bootstrappers

自动 approve CSR 请求，生成 kubelet client 证书

kubelet 建立 CSR 请求后，下一步须要建立被 approve，有两种方式：

kube-controller-manager 自动 aprrove；
手动使用命令 ·kubectl certificate approve·；

CSR 被 approve 后，kubelet 向 kube-controller-manager 请求建立 client 证书，kube-controller-manager 中的 csrapproving controller 使用 SubjectAccessReview API 来检查 kubelet 请求（对应的 group 是 system:bootstrappers）是否具备相应的权限。

建立三个 ClusterRoleBinding，分别授予 group system:bootstrappers 和 group system:nodes 进行 approve client、renew client、renew server 证书的权限（server csr 是手动 approve 的，见后文）：

cd /opt/k8s/work
cat > csr-crb.yaml <<EOF
 # Approve all CSRs for the group "system:bootstrappers"
 kind: ClusterRoleBinding
 apiVersion: rbac.authorization.k8s.io/v1
 metadata:
   name: auto-approve-csrs-for-group
 subjects:
 - kind: Group
   name: system:bootstrappers
   apiGroup: rbac.authorization.k8s.io
 roleRef:
   kind: ClusterRole
   name: system:certificates.k8s.io:certificatesigningrequests:nodeclient
   apiGroup: rbac.authorization.k8s.io
---
 # To let a node of the group "system:nodes" renew its own credentials
 kind: ClusterRoleBinding
 apiVersion: rbac.authorization.k8s.io/v1
 metadata:
   name: node-client-cert-renewal
 subjects:
 - kind: Group
   name: system:nodes
   apiGroup: rbac.authorization.k8s.io
 roleRef:
   kind: ClusterRole
   name: system:certificates.k8s.io:certificatesigningrequests:selfnodeclient
   apiGroup: rbac.authorization.k8s.io
---
# A ClusterRole which instructs the CSR approver to approve a node requesting a
# serving cert matching its client cert.
kind: ClusterRole
apiVersion: rbac.authorization.k8s.io/v1
metadata:
  name: approve-node-server-renewal-csr
rules:
- apiGroups: ["certificates.k8s.io"]
  resources: ["certificatesigningrequests/selfnodeserver"]
  verbs: ["create"]
---
 # To let a node of the group "system:nodes" renew its own server credentials
 kind: ClusterRoleBinding
 apiVersion: rbac.authorization.k8s.io/v1
 metadata:
   name: node-server-cert-renewal
 subjects:
 - kind: Group
   name: system:nodes
   apiGroup: rbac.authorization.k8s.io
 roleRef:
   kind: ClusterRole
   name: approve-node-server-renewal-csr
   apiGroup: rbac.authorization.k8s.io
EOF
kubectl apply -f csr-crb.yaml

auto-approve-csrs-for-group：自动 approve node 的第一次 CSR；注意第一次 CSR 时，请求的 Group 为 system:bootstrappers；
node-client-cert-renewal：自动 approve node 后续过时的 client 证书，自动生成的证书 Group 为 system:nodes;
node-server-cert-renewal：自动 approve node 后续过时的 server 证书，自动生成的证书 Group 为 system:nodes;

启动 kubelet 服务

source /opt/k8s/bin/environment.sh
for node_ip in ${NODE_IPS[@]}
  do
    echo ">>> ${node_ip}"
    ssh root@${node_ip} "mkdir -p ${K8S_DIR}/kubelet/kubelet-plugins/volume/exec/"
    ssh root@${node_ip} "/usr/sbin/swapoff -a"
    ssh root@${node_ip} "systemctl daemon-reload && systemctl enable kubelet && systemctl restart kubelet"
  done

启动服务前必须先建立工做目录；
关闭 swap 分区，不然 kubelet 会启动失败；

kubelet 启动后使用 --bootstrap-kubeconfig 向 kube-apiserver 发送 CSR 请求，当这个 CSR 被 approve 后，kube-controller-manager 为 kubelet 建立 TLS 客户端证书、私钥和 --kubeletconfig 文件。

注意：kube-controller-manager 须要配置 --cluster-signing-cert-file 和 --cluster-signing-key-file 参数，才会为 TLS Bootstrap 建立证书和私钥。

查看 kubelet 状况

稍等一会，三个节点的 CSR 都被自动 approved：

$ kubectl get csr
NAME        AGE   REQUESTOR                      CONDITION
csr-ct8r8   55s   system:node:zhaoyixin-k8s-02   Pending
csr-dtm97   72s   system:bootstrap:udg7tq        Approved,Issued
csr-hwsnh   70s   system:bootstrap:vl120m        Approved,Issued
csr-jxml4   57s   system:node:zhaoyixin-k8s-01   Pending
csr-tw6m6   70s   system:bootstrap:3uk8cy        Approved,Issued
csr-xh7j6   56s   system:node:zhaoyixin-k8s-03   Pending

Pending 的 CSR 用于建立 kubelet server 证书，须要手动 approve，参考后文。

全部节点均注册（NotReady 状态是预期的，后续安装了网络插件后就好）：

$ kubectl get node
NAME               STATUS     ROLES    AGE   VERSION
zhaoyixin-k8s-01   NotReady   <none>   98s   v1.16.6
zhaoyixin-k8s-02   NotReady   <none>   96s   v1.16.6
zhaoyixin-k8s-03   NotReady   <none>   96s   v1.16.6

kube-controller-manager 为各 node 生成了 kubeconfig 文件和公私钥：

$ ls -l /etc/kubernetes/kubelet.kubeconfig
-rw------- 1 root root 2258 May 30 15:08 /etc/kubernetes/kubelet.kubeconfig
$ ls -l /etc/kubernetes/cert/kubelet-client-*
-rw------- 1 root root 1289 May 30 15:08 /etc/kubernetes/cert/kubelet-client-2020-05-30-15-08-26.pem
lrwxrwxrwx 1 root root   59 May 30 15:08 /etc/kubernetes/cert/kubelet-client-current.pem -> /etc/kubernetes/cert/kubelet-client-2020-05-30-15-08-26.pem

没有自动生成 kubelet server 证书；

手动 approve server cert csr

基于安全性考虑，CSR approving controllers 不会自动 approve kubelet server 证书签名请求，须要手动 approve：

$ kubectl get csr
NAME        AGE     REQUESTOR                      CONDITION
csr-ct8r8   2m23s   system:node:zhaoyixin-k8s-02   Pending
csr-dtm97   2m40s   system:bootstrap:udg7tq        Approved,Issued
csr-hwsnh   2m38s   system:bootstrap:vl120m        Approved,Issued
csr-jxml4   2m25s   system:node:zhaoyixin-k8s-01   Pending
csr-tw6m6   2m38s   system:bootstrap:3uk8cy        Approved,Issued
csr-xh7j6   2m24s   system:node:zhaoyixin-k8s-03   Pending

$ # 手动 approve
$ kubectl get csr | grep Pending | awk '{print $1}' | xargs kubectl certificate approve

$ # 自动生成了 server 证书
$  ls -l /etc/kubernetes/cert/kubelet-*
-rw------- 1 root root 1289 May 30 15:08 /etc/kubernetes/cert/kubelet-client-2020-05-30-15-08-26.pem
lrwxrwxrwx 1 root root   59 May 30 15:08 /etc/kubernetes/cert/kubelet-client-current.pem -> /etc/kubernetes/cert/kubelet-client-2020-05-30-15-08-26.pem
-rw------- 1 root root 1338 May 30 15:12 /etc/kubernetes/cert/kubelet-server-2020-05-30-15-12-23.pem
lrwxrwxrwx 1 root root   59 May 30 15:12 /etc/kubernetes/cert/kubelet-server-current.pem -> /etc/kubernetes/cert/kubelet-server-2020-05-30-15-12-23.pem

kubelet api 认证和受权

kubelet 配置了以下认证参数：

authentication.anonymous.enabled：设置为 false，不容许匿名访问 10250 端口；
authentication.x509.clientCAFile：指定签名客户端证书的 CA 证书，开启 HTTPs 证书认证；
authentication.webhook.enabled=true：开启 HTTPs bearer token 认证；

同时配置了以下受权参数：

authroization.mode=Webhook：开启 RBAC 受权；

kubelet 收到请求后，使用 clientCAFile 对证书签名进行认证，或者查询 bearer token 是否有效。若是二者都没经过，则拒绝请求，提示 Unauthorized：

$ curl -s --cacert /etc/kubernetes/cert/ca.pem https://192.168.16.8:10250/metrics
Unauthorized

$ curl -s --cacert /etc/kubernetes/cert/ca.pem -H "Authorization: Bearer 123456" https://192.168.16.8:10250/metrics
Unauthorized

经过认证后，kubelet 使用 SubjectAccessReview API 向 kube-apiserver 发送请求，查询证书或 token 对应的 user、group 是否有操做资源的权限(RBAC)；

证书认证和受权

$ # 权限不足的证书；
$ curl -s --cacert /etc/kubernetes/cert/ca.pem --cert /etc/kubernetes/cert/kube-controller-manager.pem --key /etc/kubernetes/cert/kube-controller-manager-key.pem https://192.168.16.8:10250/metrics
Forbidden (user=system:kube-controller-manager, verb=get, resource=nodes, subresource=metrics)

$ # 使用部署 kubectl 命令行工具时建立的、具备最高权限的 admin 证书；
$ curl -s --cacert /etc/kubernetes/cert/ca.pem --cert /opt/k8s/work/admin.pem --key /opt/k8s/work/admin-key.pem https://192.168.16.8:10250/metrics|head
# HELP apiserver_audit_event_total [ALPHA] Counter of audit events generated and sent to the audit backend.
# TYPE apiserver_audit_event_total counter
apiserver_audit_event_total 0
# HELP apiserver_audit_requests_rejected_total [ALPHA] Counter of apiserver requests rejected due to an error in audit logging backend.
# TYPE apiserver_audit_requests_rejected_total counter
apiserver_audit_requests_rejected_total 0
# HELP apiserver_client_certificate_expiration_seconds [ALPHA] Distribution of the remaining lifetime on the certificate used to authenticate a request.
# TYPE apiserver_client_certificate_expiration_seconds histogram
apiserver_client_certificate_expiration_seconds_bucket{le="0"} 0
apiserver_client_certificate_expiration_seconds_bucket{le="1800"} 0

--cacert、--cert、--key 的参数值必须是文件路径，如上面的 ./admin.pem 不能省略 ./，不然返回 401 Unauthorized；

bear token 认证和受权

建立一个 ServiceAccount，将它和 ClusterRole system:kubelet-api-admin 绑定，从而具备调用 kubelet API 的权限：

kubectl create sa kubelet-api-test
kubectl create clusterrolebinding kubelet-api-test --clusterrole=system:kubelet-api-admin --serviceaccount=default:kubelet-api-test
SECRET=$(kubectl get secrets | grep kubelet-api-test | awk '{print $1}')
TOKEN=$(kubectl describe secret ${SECRET} | grep -E '^token' | awk '{print $2}')
echo ${TOKEN}

$ curl -s --cacert /etc/kubernetes/cert/ca.pem -H "Authorization: Bearer ${TOKEN}" https://192.168.16.8:10250/metrics | head
# HELP apiserver_audit_event_total [ALPHA] Counter of audit events generated and sent to the audit backend.
# TYPE apiserver_audit_event_total counter
apiserver_audit_event_total 0
# HELP apiserver_audit_requests_rejected_total [ALPHA] Counter of apiserver requests rejected due to an error in audit logging backend.
# TYPE apiserver_audit_requests_rejected_total counter
apiserver_audit_requests_rejected_total 0
# HELP apiserver_client_certificate_expiration_seconds [ALPHA] Distribution of the remaining lifetime on the certificate used to authenticate a request.
# TYPE apiserver_client_certificate_expiration_seconds histogram
apiserver_client_certificate_expiration_seconds_bucket{le="0"} 0
apiserver_client_certificate_expiration_seconds_bucket{le="1800"} 0

cadvisor 和 metrics

cadvisor 是内嵌在 kubelet 二进制中的，统计所在节点各容器的资源(CPU、内存、磁盘、网卡)使用状况的服务。

浏览器访问 kube-apiserver 的安全端口 6443 时，提示证书不被信任。这是由于 kube-apiserver 的 server 证书是咱们建立的根证书 ca.pem 签名的，须要将根证书 ca.pem 导入操做系统，并设置永久信任。

对于 windows 系统使用如下命令导入 ca.perm：

keytool -import -v -trustcacerts -alias appmanagement -file "PATH...\\ca.pem" -storepass password -keystore cacerts

咱们须要给浏览器生成一个 client 证书，访问 apiserver 的 6443 https 端口时使用。

这里使用部署 kubectl 命令行工具时建立的 admin 证书、私钥和上面的 ca 证书，建立一个浏览器可使用 PKCS#12/PFX 格式的证书：

$ openssl pkcs12 -export -out admin.pfx -inkey admin-key.pem -in admin.pem -certfile ca.pem

将建立的 admin.pfx 导入到系统的证书中便可。

浏览器访问 https://192.168.16.8:10250/metrics 和 https://192.168.16.8:10250/metrics/cadvisor 分别返回 kubelet 和 cadvisor 的 metrics。

注意：

kubelet.config.json 设置 authentication.anonymous.enabled 为 false，不容许匿名证书访问 10250 的 https 服务。

客户端选择证书的原理

证书选择是在客户端和服务端 SSL/TLS 握手协商阶段商定的；
服务端若是要求客户端提供证书，则在握手时会向客户端发送一个它接受的 CA 列表；
客户端查找它的证书列表(通常是操做系统的证书，对于 Mac 为 keychain)，看有没有被 CA 签名的证书，若是有，则将它们提供给用户选择（证书的私钥）；
用户选择一个证书私钥，而后客户端将使用它和服务端通讯；

4.部署 kube-proxy 组件

kube-proxy 运行在全部 worker 节点上，它监听 apiserver 中 service 和 endpoint 的变化状况，建立路由规则以提供服务 IP 和负载均衡功能。

下载和分发 kube-proxy 二进制文件

参考【k8s部署】5. 部署 master 节点中的《下载二进制文件》一节。

建立 kube-proxy 证书

建立证书签名请求：

cd /opt/k8s/work
cat > kube-proxy-csr.json <<EOF
{
  "CN": "system:kube-proxy",
  "key": {
    "algo": "rsa",
    "size": 2048
  },
  "names": [
    {
      "C": "CN",
      "ST": "BeiJing",
      "L": "BeiJing",
      "O": "k8s",
      "OU": "zhaoyixin"
    }
  ]
}
EOF

CN：指定该证书的 User 为 system:kube-proxy；
预约义的 RoleBinding system:node-proxier 将 User system:kube-proxy 与 Role system:node-proxier 绑定，该 Role 授予了调用 kube-apiserver Proxy 相关 API 的权限；
该证书只会被 kube-proxy 当作 client 证书使用，因此 hosts 字段为空；

生成证书和私钥：

cd /opt/k8s/work
cfssl gencert -ca=/opt/k8s/work/ca.pem \
  -ca-key=/opt/k8s/work/ca-key.pem \
  -config=/opt/k8s/work/ca-config.json \
  -profile=kubernetes  kube-proxy-csr.json | cfssljson -bare kube-proxy
ls kube-proxy*

建立和分发 kubeconfig 文件

cd /opt/k8s/work
source /opt/k8s/bin/environment.sh
kubectl config set-cluster kubernetes \
  --certificate-authority=/opt/k8s/work/ca.pem \
  --embed-certs=true \
  --server=${KUBE_APISERVER} \
  --kubeconfig=kube-proxy.kubeconfig

kubectl config set-credentials kube-proxy \
  --client-certificate=kube-proxy.pem \
  --client-key=kube-proxy-key.pem \
  --embed-certs=true \
  --kubeconfig=kube-proxy.kubeconfig

kubectl config set-context default \
  --cluster=kubernetes \
  --user=kube-proxy \
  --kubeconfig=kube-proxy.kubeconfig

kubectl config use-context default --kubeconfig=kube-proxy.kubeconfig

分发 kubeconfig 文件：

cd /opt/k8s/work
source /opt/k8s/bin/environment.sh
for node_name in ${NODE_NAMES[@]}
  do
    echo ">>> ${node_name}"
    scp kube-proxy.kubeconfig root@${node_name}:/etc/kubernetes/
  done

建立 kube-proxy 配置文件

建立 kube-proxy config 文件模板：

cd /opt/k8s/work
source /opt/k8s/bin/environment.sh
cat > kube-proxy-config.yaml.template <<EOF
kind: KubeProxyConfiguration
apiVersion: kubeproxy.config.k8s.io/v1alpha1
clientConnection:
  burst: 200
  kubeconfig: "/etc/kubernetes/kube-proxy.kubeconfig"
  qps: 100
bindAddress: ##NODE_IP##
healthzBindAddress: ##NODE_IP##:10256
metricsBindAddress: ##NODE_IP##:10249
enableProfiling: true
clusterCIDR: ${CLUSTER_CIDR}
hostnameOverride: ##NODE_NAME##
mode: "ipvs"
portRange: ""
iptables:
  masqueradeAll: false
ipvs:
  scheduler: rr
  excludeCIDRs: []
EOF

bindAddress: 监听地址；
clientConnection.kubeconfig: 链接 apiserver 的 kubeconfig 文件；
clusterCIDR: kube-proxy 根据 --cluster-cidr 判断集群内部和外部流量，指定 --cluster-cidr 或 --masquerade-all 选项后 kube-proxy 才会对访问 Service IP 的请求作 SNAT；
hostnameOverride: 参数值必须与 kubelet 的值一致，不然 kube-proxy 启动后会找不到该 Node，从而不会建立任何 ipvs 规则；
mode: 使用 ipvs 模式；

为各节点建立和分发 kube-proxy 配置文件：

cd /opt/k8s/work
source /opt/k8s/bin/environment.sh
for (( i=0; i < 3; i++ ))
  do 
    echo ">>> ${NODE_NAMES[i]}"
    sed -e "s/##NODE_NAME##/${NODE_NAMES[i]}/" -e "s/##NODE_IP##/${NODE_IPS[i]}/" kube-proxy-config.yaml.template > kube-proxy-config-${NODE_NAMES[i]}.yaml.template
    scp kube-proxy-config-${NODE_NAMES[i]}.yaml.template root@${NODE_NAMES[i]}:/etc/kubernetes/kube-proxy-config.yaml
  done

建立和分发 kube-proxy systemd unit 文件

cd /opt/k8s/work
source /opt/k8s/bin/environment.sh
cat > kube-proxy.service <<EOF
[Unit]
Description=Kubernetes Kube-Proxy Server
Documentation=https://github.com/GoogleCloudPlatform/kubernetes
After=network.target

[Service]
WorkingDirectory=${K8S_DIR}/kube-proxy
ExecStart=/opt/k8s/bin/kube-proxy \\
  --config=/etc/kubernetes/kube-proxy-config.yaml \\
  --logtostderr=true \\
  --v=2
Restart=on-failure
RestartSec=5
LimitNOFILE=65536

[Install]
WantedBy=multi-user.target
EOF

分发 kube-proxy systemd unit 文件：

cd /opt/k8s/work
source /opt/k8s/bin/environment.sh
for node_name in ${NODE_NAMES[@]}
  do 
    echo ">>> ${node_name}"
    scp kube-proxy.service root@${node_name}:/etc/systemd/system/
  done

启动并检查 kube-proxy 服务

cd /opt/k8s/work
source /opt/k8s/bin/environment.sh
for node_ip in ${NODE_IPS[@]}
  do
    echo ">>> ${node_ip}"
    ssh root@${node_ip} "mkdir -p ${K8S_DIR}/kube-proxy"
    ssh root@${node_ip} "modprobe ip_vs_rr"
    ssh root@${node_ip} "systemctl daemon-reload && systemctl enable kube-proxy && systemctl restart kube-proxy"
  done

检查启动结果

source /opt/k8s/bin/environment.sh
for node_ip in ${NODE_IPS[@]}
  do
    echo ">>> ${node_ip}"
    ssh root@${node_ip} "systemctl status kube-proxy|grep Active"
  done

确保状态为 active (running)，不然经过 journalctl -u kube-proxy 查看日志，确认缘由。

查看监听端口

$ sudo netstat -lnpt|grep kube-prox
tcp        0      0 192.168.16.8:10249      0.0.0.0:*               LISTEN      11115/kube-proxy    
tcp        0      0 192.168.16.8:10256      0.0.0.0:*               LISTEN      11115/kube-proxy

10249：http prometheus metrics port;
10256：http healthz port;

查看 ipvs 路由规则

source /opt/k8s/bin/environment.sh
for node_ip in ${NODE_IPS[@]}
  do
    echo ">>> ${node_ip}"
    ssh root@${node_ip} "/usr/sbin/ipvsadm -ln"
  done

预期输出：

>>> 192.168.16.8
IP Virtual Server version 1.2.1 (size=4096)
Prot LocalAddress:Port Scheduler Flags
  -> RemoteAddress:Port           Forward Weight ActiveConn InActConn
TCP  10.254.0.1:443 rr
  -> 192.168.16.6:6443            Masq    1      0          0         
  -> 192.168.16.8:6443            Masq    1      0          0         
  -> 192.168.16.10:6443           Masq    1      0          0         
>>> 192.168.16.10
IP Virtual Server version 1.2.1 (size=4096)
Prot LocalAddress:Port Scheduler Flags
  -> RemoteAddress:Port           Forward Weight ActiveConn InActConn
TCP  10.254.0.1:443 rr
  -> 192.168.16.6:6443            Masq    1      0          0         
  -> 192.168.16.8:6443            Masq    1      0          0         
  -> 192.168.16.10:6443           Masq    1      0          0         
>>> 192.168.16.6
IP Virtual Server version 1.2.1 (size=4096)
Prot LocalAddress:Port Scheduler Flags
  -> RemoteAddress:Port           Forward Weight ActiveConn InActConn
TCP  10.254.0.1:443 rr
  -> 192.168.16.6:6443            Masq    1      0          0         
  -> 192.168.16.8:6443            Masq    1      0          0         
  -> 192.168.16.10:6443           Masq    1      0          0

可见全部经过 https 访问 K8S SVC kubernetes 的请求都转发到 kube-apiserver 节点的 6443 端口。

5.部署 calico 网络

kubernetes 要求集群内各节点(包括 master 节点)能经过 Pod 网段互联互通。

calico 使用 IPIP 或 BGP 技术（默认为 IPIP）为各节点建立一个能够互通的 Pod 网络。

安装 calico 网络插件

cd /opt/k8s/work
curl https://docs.projectcalico.org/manifests/calico.yaml -O

修改配置：

$ cp calico.yaml calico.yaml.orig
$ diff calico.yaml.orig calico.yaml
630c630,632
<               value: "192.168.0.0/16"
---
>               value: "172.30.0.0/16"
>             - name: IP_AUTODETECTION_METHOD
>               value: "interface=eth.*"
699c701
<             path: /opt/cni/bin
---
>             path: /opt/k8s/bin

将 Pod 网段地址修改成 172.30.0.0/16，即在【k8s部署】1. 环境准备和初始化文中 environment.sh 环境变量文件中定义的 Pod 网段范围；
calico 自动探查互联网卡，若是有多快网卡，则能够配置用于互联的网络接口命名正则表达式，如上面的 eth.*(根据本身服务器的网络接口名修改)；

运行 calico 插件：

$ kubectl apply -f  calico.yaml

calico 插件以 daemonset 方式运行在全部的 K8S 节点上。

查看 calico 运行状态

$ kubectl get pods -n kube-system -o wide
NAME                                       READY   STATUS    RESTARTS   AGE   IP              NODE               NOMINATED NODE   READINESS GATES
calico-kube-controllers-77d6cbc65f-c9s88   1/1     Running   0          22m   172.30.219.1    zhaoyixin-k8s-02   <none>           <none>
calico-node-gf8h2                          1/1     Running   0          22m   192.168.16.10   zhaoyixin-k8s-02   <none>           <none>
calico-node-n26rj                          1/1     Running   0          22m   192.168.16.8    zhaoyixin-k8s-01   <none>           <none>
calico-node-nx8hz                          1/1     Running   0          22m   192.168.16.6    zhaoyixin-k8s-03   <none>           <none>

使用 crictl 命令查看 calico 使用的镜像：

$ crictl  images
docker.io/calico/cni                                      v3.14.1             35a7136bc71a7       77.6MB
docker.io/calico/node                                     v3.14.1             04a9b816c7535       90.6MB
docker.io/calico/pod2daemon-flexvol                       v3.14.1             7f93af2e7e114       37.5MB
registry.cn-beijing.aliyuncs.com/images_k8s/pause-amd64   3.1                 21a595adc69ca       326kB

若是 crictl 输出为空或执行失败，则有多是缺乏配置文件 /etc/crictl.yaml 致使的，该文件的配置以下：

$ cat /etc/crictl.yaml
runtime-endpoint: unix:///run/containerd/containerd.sock
image-endpoint: unix:///run/containerd/containerd.sock
timeout: 10
debug: false

参考

opsnull/follow-me-install-kubernetes-cluster