使用kubeadm部署Kubernetes 1.28
2023-08-20
kubeadm是Kubernetes官方提供的用于快速安部署Kubernetes集群的工具。伴随Kubernetes每个版本的发布，kubeadm可能会对集群配置方面的一些实践做调整，通过实验kubeadm可以学习到Kubernetes官方在集群配置上一些新的最佳实践。
1.准备 #
1.1 系统配置 #
在安装之前，需要先做好如下准备。
3台Linux主机如下：
- node4 - Ubuntu 22.04
- node5 - openEuler release 22.03 (LTS-SP2)
- node6 - Rocky Linux release 8.8 (Green Obsidian)
1cat /etc/hosts
2192.168.96.154 node4
3192.168.96.155 node5
4192.168.96.156 node6
在各个主机上完成下面的系统配置。
如果各个主机启用了防火墙策略，需要开放Kubernetes各个组件所需要的端口，可以查看Ports and Protocols中的内容, 开放相关端口或者关闭主机的防火墙。
创建/etc/modules-load.d/containerd.conf配置文件，确保在系统启动时自动加载所需的内核模块，以满足容器运行时的要求:
1cat << EOF > /etc/modules-load.d/containerd.conf
2overlay
3br_netfilter
4EOF
执行以下命令使配置生效:
1modprobe overlay
2modprobe br_netfilter
创建/etc/sysctl.d/99-kubernetes-cri.conf配置文件：
1cat << EOF > /etc/sysctl.d/99-kubernetes-cri.conf
2net.bridge.bridge-nf-call-ip6tables = 1
3net.bridge.bridge-nf-call-iptables = 1
4net.ipv4.ip_forward = 1
5user.max_user_namespaces=28633
6EOF
执行以下命令使配置生效:
1sysctl -p /etc/sysctl.d/99-kubernetes-cri.conf
在文件名
/etc/sysctl.d/99-kubernetes-cri.conf中，“99” 代表文件的优先级或顺序。sysctl是Linux内核参数的配置工具，它可以通过修改/proc/sys/目录下的文件来设置内核参数。在/etc/sysctl.d/目录中，可以放置一系列的配置文件，以便在系统启动时自动加载这些参数。这些配置文件按照文件名的字母顺序逐个加载。数字前缀用于指定加载的顺序，较小的数字表示较高的优先级。
1.2 配置服务器支持开启ipvs的前提条件 #
由于ipvs已经加入到了内核的主干，所以为kube-proxy开启ipvs的前提需要加载以下的内核模块：
1ip_vs
2ip_vs_rr
3ip_vs_wrr
4ip_vs_sh
5nf_conntrack_ipv4
创建/etc/modules-load.d/ipvs.conf文件，保证在节点重启后能自动加载所需模块:
1cat > /etc/modules-load.d/ipvs.conf <<EOF
2ip_vs
3ip_vs_rr
4ip_vs_wrr
5ip_vs_sh
6EOF
执行以下命令使配置立即生效:
1modprobe ip_vs
2modprobe ip_vs_rr
3modprobe ip_vs_wrr
4modprobe ip_vs_sh
使用lsmod | grep -e ip_vs -e nf_conntrack命令查看是否已经正确加载所需的内核模块。
接下来还需要确保各个节点上已经安装了ipset软件包，为了便于查看ipvs的代理规则，最好安装一下管理工具ipvsadm。
在Ubuntu系统上执行:
1apt install -y ipset ipvsadm
在openEuler或Rocky Linux系统上执行:
1yum install -y ipset ipvsadm
如果不满足以上前提条件，即使kube-proxy的配置开启了ipvs模式，也会退回到iptables模式。
1.3 部署容器运行时Containerd #
在各个服务器节点上安装容器运行时Containerd。
下载Containerd的二进制包， 需要注意cri-containerd-(cni-)-VERSION-OS-ARCH.tar.gz发行包自containerd 1.6版本起已经被弃用，在某些 Linux 发行版上无法正常工作，并将在containerd 2.0版本中移除，这里下载containerd-<VERSION>-<OS>-<ARCH>.tar.gz的发行包，后边在单独下载安装runc和CNI plugins:
1wget https://github.com/containerd/containerd/releases/download/v1.7.3/containerd-1.7.3-linux-amd64.tar.gz
将其解压缩到/usr/local下:
1tar Cxzvf /usr/local containerd-1.7.3-linux-amd64.tar.gz
2
3bin/
4bin/containerd-shim-runc-v1
5bin/containerd-shim-runc-v2
6bin/containerd-stress
7bin/containerd
8bin/containerd-shim
9bin/ctr
接下来从runc的github上单独下载安装runc，该二进制文件是静态构建的，并且应该适用于任何Linux发行版。
1wget https://github.com/opencontainers/runc/releases/download/v1.1.9/runc.amd64
2install -m 755 runc.amd64 /usr/local/sbin/runc
接下来生成containerd的配置文件:
1mkdir -p /etc/containerd
2containerd config default > /etc/containerd/config.toml
根据文档Container runtimes中的内容，对于使用systemd作为init system的Linux的发行版，使用systemd作为容器的cgroup driver可以确保服务器节点在资源紧张的情况更加稳定，因此这里配置各个节点上containerd的cgroup driver为systemd。
修改前面生成的配置文件/etc/containerd/config.toml：
1[plugins."io.containerd.grpc.v1.cri".containerd.runtimes.runc]
2 ...
3 [plugins."io.containerd.grpc.v1.cri".containerd.runtimes.runc.options]
4 SystemdCgroup = true
再修改/etc/containerd/config.toml中的
1[plugins."io.containerd.grpc.v1.cri"]
2 ...
3 # sandbox_image = "registry.k8s.io/pause:3.8"
4 sandbox_image = "registry.aliyuncs.com/google_containers/pause:3.9"
为了通过systemd启动containerd，请还需要从https://raw.githubusercontent.com/containerd/containerd/main/containerd.service下载containerd.service单元文件，并将其放置在 /etc/systemd/system/containerd.service中。
1cat << EOF > /etc/systemd/system/containerd.service
2[Unit]
3Description=containerd container runtime
4Documentation=https://containerd.io
5After=network.target local-fs.target
6
7[Service]
8#uncomment to enable the experimental sbservice (sandboxed) version of containerd/cri integration
9#Environment="ENABLE_CRI_SANDBOXES=sandboxed"
10ExecStartPre=-/sbin/modprobe overlay
11ExecStart=/usr/local/bin/containerd
12
13Type=notify
14Delegate=yes
15KillMode=process
16Restart=always
17RestartSec=5
18# Having non-zero Limit*s causes performance problems due to accounting overhead
19# in the kernel. We recommend using cgroups to do container-local accounting.
20LimitNPROC=infinity
21LimitCORE=infinity
22LimitNOFILE=infinity
23# Comment TasksMax if your systemd version does not supports it.
24# Only systemd 226 and above support this version.
25TasksMax=infinity
26OOMScoreAdjust=-999
27
28[Install]
29WantedBy=multi-user.target
30EOF
配置containerd开机启动，并启动containerd，执行以下命令:
1systemctl daemon-reload
2systemctl enable containerd --now
3systemctl status containerd
下载安装crictl工具:
1wget https://github.com/kubernetes-sigs/cri-tools/releases/download/v1.28.0/crictl-v1.28.0-linux-amd64.tar.gz
2tar -zxvf crictl-v1.28.0-linux-amd64.tar.gz
3install -m 755 crictl /usr/local/bin/crictl
使用crictl测试一下，确保可以打印出版本信息并且没有错误信息输出:
1crictl --runtime-endpoint=unix:///run/containerd/containerd.sock version
2
3Version: 0.1.0
4RuntimeName: containerd
5RuntimeVersion: v1.7.3
6RuntimeApiVersion: v1
2.使用kubeadm部署Kubernetes #
2.1 安装kubeadm和kubelet #
下面在各节点安装kubeadm和kubelet：
在Ubuntu系统上执行下面的命令:
1apt-get update
2apt-get install -y apt-transport-https ca-certificates curl
3
4curl -s https://mirrors.aliyun.com/kubernetes/apt/doc/apt-key.gpg | sudo apt-key add -
5
6
7tee /etc/apt/sources.list.d/kubernetes.list <<-'EOF'
8deb https://mirrors.aliyun.com/kubernetes/apt/ kubernetes-xenial main
9EOF
10
11apt-get update
12
13apt install kubelet kubeadm kubectl
14
15apt-mark hold kubelet kubeadm kubectl
在openEuler和Rocky Linux系统中执行下面的命令:
1cat <<EOF > /etc/yum.repos.d/kubernetes.repo
2[kubernetes]
3name=Kubernetes
4baseurl=http://mirrors.aliyun.com/kubernetes/yum/repos/kubernetes-el7-x86_64
5enabled=1
6gpgcheck=1
7repo_gpgcheck=0
8gpgkey=http://mirrors.aliyun.com/kubernetes/yum/doc/yum-key.gpg
9 http://mirrors.aliyun.com/kubernetes/yum/doc/rpm-package-key.gpg
10EOF
11
12
13yum makecache
14yum install kubelet kubeadm kubectl
上面的命令在安装kubeadm, kubectl, kubelet 时，会自动安装依赖项conntrack, cri-tools, ebtables, kubernetes-cni, soca
运行kubelet --help可以看到原来kubelet的绝大多数命令行flag参数都被DEPRECATED了，官方推荐我们使用--config指定配置文件，并在配置文件中指定原来这些flag所配置的内容。具体内容可以查看这里Set Kubelet parameters via a config file。最初Kubernetes这么做是为了支持动态Kubelet配置（Dynamic Kubelet Configuration），但动态Kubelet配置特性从k8s 1.22中已弃用，并在1.24中被移除。如果需要调整集群汇总所有节点kubelet的配置，还是推荐使用ansible等工具将配置分发到各个节点。
kubelet的配置文件必须是json或yaml格式，具体可查看这里。
Kubernetes 1.8开始要求关闭系统的Swap，如果不关闭，默认配置下kubelet将无法启动。 关闭系统的Swap方法如下:
1swapoff -a
修改/etc/fstab文件，注释掉 SWAP 的自动挂载，使用free -m确认swap已经关闭。
swappiness参数调整，修改/etc/sysctl.d/99-kubernetes-cri.conf添加下面一行：
1vm.swappiness=0
执行sysctl -p /etc/sysctl.d/99-kubernetes-cri.conf使修改生效。
2.2 使用kubeadm init初始化集群 #
在各节点开机启动kubelet服务：
1systemctl enable kubelet.service
使用kubeadm config print init-defaults --component-configs KubeletConfiguration可以打印集群初始化默认的使用的配置：
1apiVersion: kubeadm.k8s.io/v1beta3
2bootstrapTokens:
3- groups:
4 - system:bootstrappers:kubeadm:default-node-token
5 token: abcdef.0123456789abcdef
6 ttl: 24h0m0s
7 usages:
8 - signing
9 - authentication
10kind: InitConfiguration
11localAPIEndpoint:
12 advertiseAddress: 1.2.3.4
13 bindPort: 6443
14nodeRegistration:
15 criSocket: unix:///var/run/containerd/containerd.sock
16 imagePullPolicy: IfNotPresent
17 name: node
18 taints: null
19---
20apiServer:
21 timeoutForControlPlane: 4m0s
22apiVersion: kubeadm.k8s.io/v1beta3
23certificatesDir: /etc/kubernetes/pki
24clusterName: kubernetes
25controllerManager: {}
26dns: {}
27etcd:
28 local:
29 dataDir: /var/lib/etcd
30imageRepository: registry.k8s.io
31kind: ClusterConfiguration
32kubernetesVersion: 1.28.0
33networking:
34 dnsDomain: cluster.local
35 serviceSubnet: 10.96.0.0/12
36scheduler: {}
37---
38apiVersion: kubelet.config.k8s.io/v1beta1
39authentication:
40 anonymous:
41 enabled: false
42 webhook:
43 cacheTTL: 0s
44 enabled: true
45 x509:
46 clientCAFile: /etc/kubernetes/pki/ca.crt
47authorization:
48 mode: Webhook
49 webhook:
50 cacheAuthorizedTTL: 0s
51 cacheUnauthorizedTTL: 0s
52cgroupDriver: systemd
53clusterDNS:
54- 10.96.0.10
55clusterDomain: cluster.local
56containerRuntimeEndpoint: ""
57cpuManagerReconcilePeriod: 0s
58evictionPressureTransitionPeriod: 0s
59fileCheckFrequency: 0s
60healthzBindAddress: 127.0.0.1
61healthzPort: 10248
62httpCheckFrequency: 0s
63imageMinimumGCAge: 0s
64kind: KubeletConfiguration
65logging:
66 flushFrequency: 0
67 options:
68 json:
69 infoBufferSize: "0"
70 verbosity: 0
71memorySwap: {}
72nodeStatusReportFrequency: 0s
73nodeStatusUpdateFrequency: 0s
74resolvConf: /run/systemd/resolve/resolv.conf
75rotateCertificates: true
76runtimeRequestTimeout: 0s
77shutdownGracePeriod: 0s
78shutdownGracePeriodCriticalPods: 0s
79staticPodPath: /etc/kubernetes/manifests
80streamingConnectionIdleTimeout: 0s
81syncFrequency: 0s
82volumeStatsAggPeriod: 0s
从默认的配置中可以看到，可以使用imageRepository定制在集群初始化时拉取k8s所需镜像的地址。基于默认配置定制出本次使用kubeadm初始化集群所需的配置文件kubeadm.yaml：
1apiVersion: kubeadm.k8s.io/v1beta3
2kind: InitConfiguration
3localAPIEndpoint:
4 advertiseAddress: 192.168.96.154
5 bindPort: 6443
6nodeRegistration:
7 criSocket: unix:///run/containerd/containerd.sock
8 taints:
9 - effect: PreferNoSchedule
10 key: node-role.kubernetes.io/master
11---
12apiVersion: kubeadm.k8s.io/v1beta3
13kind: ClusterConfiguration
14kubernetesVersion: 1.28.0
15imageRepository: registry.aliyuncs.com/google_containers
16networking:
17 podSubnet: 10.244.0.0/16
18---
19apiVersion: kubelet.config.k8s.io/v1beta1
20kind: KubeletConfiguration
21cgroupDriver: systemd
22failSwapOn: false
23---
24apiVersion: kubeproxy.config.k8s.io/v1alpha1
25kind: KubeProxyConfiguration
26mode: ipvs
这里定制了imageRepository为阿里云的registry，避免因gcr被墙，无法直接拉取镜像。criSocket设置了容器运行时为containerd。
同时设置kubelet的cgroupDriver为systemd，设置kube-proxy代理模式为ipvs。
在开始初始化集群之前可以使用kubeadm config images pull --config kubeadm.yaml预先在各个服务器节点上拉取所k8s需要的容器镜像。
1kubeadm config images pull --config kubeadm.yaml
2
3[config/images] Pulled registry.aliyuncs.com/google_containers/kube-apiserver:v1.28.0
4[config/images] Pulled registry.aliyuncs.com/google_containers/kube-controller-manager:v1.28.0
5[config/images] Pulled registry.aliyuncs.com/google_containers/kube-scheduler:v1.28.0
6[config/images] Pulled registry.aliyuncs.com/google_containers/kube-proxy:v1.28.0
7[config/images] Pulled registry.aliyuncs.com/google_containers/pause:3.9
8[config/images] Pulled registry.aliyuncs.com/google_containers/etcd:3.5.9-0
9[config/images] Pulled registry.aliyuncs.com/google_containers/coredns:v1.10.1
接下来使用kubeadm初始化集群，选择node4作为Master Node，在node4上执行下面的命令：
1kubeadm init --config kubeadm.yaml
2
3[init] Using Kubernetes version: v1.28.0
4[preflight] Running pre-flight checks
5 [WARNING Swap]: swap is enabled; production deployments should disable swap unless testing the NodeSwap feature gate of the kubelet
6[preflight] Pulling images required for setting up a Kubernetes cluster
7[preflight] This might take a minute or two, depending on the speed of your internet connection
8[preflight] You can also perform this action in beforehand using 'kubeadm config images pull'
9[certs] Using certificateDir folder "/etc/kubernetes/pki"
10[certs] Generating "ca" certificate and key
11[certs] Generating "apiserver" certificate and key
12[certs] apiserver serving cert is signed for DNS names [kubernetes kubernetes.default kubernetes.default.svc kubernetes.default.svc.cluster.local node4] and IPs [10.96.0.1 192.168.96.154]
13[certs] Generating "apiserver-kubelet-client" certificate and key
14[certs] Generating "front-proxy-ca" certificate and key
15[certs] Generating "front-proxy-client" certificate and key
16[certs] Generating "etcd/ca" certificate and key
17[certs] Generating "etcd/server" certificate and key
18[certs] etcd/server serving cert is signed for DNS names [localhost node4] and IPs [192.168.96.154 127.0.0.1 ::1]
19[certs] Generating "etcd/peer" certificate and key
20[certs] etcd/peer serving cert is signed for DNS names [localhost node4] and IPs [192.168.96.154 127.0.0.1 ::1]
21[certs] Generating "etcd/healthcheck-client" certificate and key
22[certs] Generating "apiserver-etcd-client" certificate and key
23[certs] Generating "sa" key and public key
24[kubeconfig] Using kubeconfig folder "/etc/kubernetes"
25[kubeconfig] Writing "admin.conf" kubeconfig file
26[kubeconfig] Writing "kubelet.conf" kubeconfig file
27[kubeconfig] Writing "controller-manager.conf" kubeconfig file
28[kubeconfig] Writing "scheduler.conf" kubeconfig file
29[etcd] Creating static Pod manifest for local etcd in "/etc/kubernetes/manifests"
30[control-plane] Using manifest folder "/etc/kubernetes/manifests"
31[control-plane] Creating static Pod manifest for "kube-apiserver"
32[control-plane] Creating static Pod manifest for "kube-controller-manager"
33[control-plane] Creating static Pod manifest for "kube-scheduler"
34[kubelet-start] Writing kubelet environment file with flags to file "/var/lib/kubelet/kubeadm-flags.env"
35[kubelet-start] Writing kubelet configuration to file "/var/lib/kubelet/config.yaml"
36[kubelet-start] Starting the kubelet
37[wait-control-plane] Waiting for the kubelet to boot up the control plane as static Pods from directory "/etc/kubernetes/manifests". This can take up to 4m0s
38[apiclient] All control plane components are healthy after 17.507599 seconds
39[upload-config] Storing the configuration used in ConfigMap "kubeadm-config" in the "kube-system" Namespace
40[kubelet] Creating a ConfigMap "kubelet-config" in namespace kube-system with the configuration for the kubelets in the cluster
41[upload-certs] Skipping phase. Please see --upload-certs
42[mark-control-plane] Marking the node node4 as control-plane by adding the labels: [node-role.kubernetes.io/control-plane node.kubernetes.io/exclude-from-external-load-balancers]
43[mark-control-plane] Marking the node node4 as control-plane by adding the taints [node-role.kubernetes.io/master:PreferNoSchedule]
44[bootstrap-token] Using token: ag6egz.xjq1zz01meq8iboq
45[bootstrap-token] Configuring bootstrap tokens, cluster-info ConfigMap, RBAC Roles
46[bootstrap-token] Configured RBAC rules to allow Node Bootstrap tokens to get nodes
47[bootstrap-token] Configured RBAC rules to allow Node Bootstrap tokens to post CSRs in order for nodes to get long term certificate credentials
48[bootstrap-token] Configured RBAC rules to allow the csrapprover controller automatically approve CSRs from a Node Bootstrap Token
49[bootstrap-token] Configured RBAC rules to allow certificate rotation for all node client certificates in the cluster
50[bootstrap-token] Creating the "cluster-info" ConfigMap in the "kube-public" namespace
51[kubelet-finalize] Updating "/etc/kubernetes/kubelet.conf" to point to a rotatable kubelet client certificate and key
52[addons] Applied essential addon: CoreDNS
53[addons] Applied essential addon: kube-proxy
54
55Your Kubernetes control-plane has initialized successfully!
56
57To start using your cluster, you need to run the following as a regular user:
58
59 mkdir -p $HOME/.kube
60 sudo cp -i /etc/kubernetes/admin.conf $HOME/.kube/config
61 sudo chown $(id -u):$(id -g) $HOME/.kube/config
62
63Alternatively, if you are the root user, you can run:
64
65 export KUBECONFIG=/etc/kubernetes/admin.conf
66
67You should now deploy a pod network to the cluster.
68Run "kubectl apply -f [podnetwork].yaml" with one of the options listed at:
69 https://kubernetes.io/docs/concepts/cluster-administration/addons/
70
71Then you can join any number of worker nodes by running the following on each as root:
72
73kubeadm join 192.168.96.154:6443 --token ag6egz.xjq1zz01meq8iboq \
74 --discovery-token-ca-cert-hash sha256:3a13ba07a146b904a10fe2e3f0ea0056890f522c327eed073f8952a5b182883a
上面记录了完成的初始化输出的内容，根据输出的内容基本上可以看出手动初始化安装一个Kubernetes集群所需要的关键步骤。 其中有以下关键内容：
[certs]生成相关的各种证书[kubeconfig]生成相关的kubeconfig文件[kubelet-start]生成kubelet的配置文件"/var/lib/kubelet/config.yaml"[control-plane]使用/etc/kubernetes/manifests目录中的yaml文件创建apiserver、controller-manager、scheduler的静态pod[bootstraptoken]生成token记录下来，后边使用kubeadm join往集群中添加节点时会用到[addons]安装基本插件:CoreDNS,kube-proxy- 下面的命令是配置常规用户如何使用kubectl访问集群：
1mkdir -p $HOME/.kube 2sudo cp -i /etc/kubernetes/admin.conf $HOME/.kube/config 3sudo chown $(id -u):$(id -g) $HOME/.kube/config - 最后给出了将另外2个节点加入集群的命令:
1kubeadm join 192.168.96.154:6443 --token ag6egz.xjq1zz01meq8iboq \
2 --discovery-token-ca-cert-hash sha256:3a13ba07a146b904a10fe2e3f0ea0056890f522c327eed073f8952a5b182883a
查看一下集群状态，确认个组件都处于healthy状态，结果出现了错误:
1kubectl get cs
2
3Warning: v1 ComponentStatus is deprecated in v1.19+
4NAME STATUS MESSAGE ERROR
5controller-manager Healthy ok
6scheduler Healthy ok
7etcd-0 Healthy ok
集群初始化如果遇到问题，可以使用kubeadm reset命令进行清理。
2.3 安装包管理器helm 3 #
Helm是Kubernetes的包管理器，后续流程也将使用Helm安装Kubernetes的常用组件。 这里先在master节点node4上安装helm。
1wget https://get.helm.sh/helm-v3.12.3-linux-amd64.tar.gz
2tar -zxvf helm-v3.12.3-linux-amd64.tar.gz
3install -m 755 linux-amd64/helm /usr/local/bin/helm
执行helm list确认没有错误输出。
2.4 部署Pod Network组件Calico #
选择calico作为k8s的Pod网络组件，下面使用helm在k8s集群中安装calico。
下载tigera-operator的helm chart:
1wget https://github.com/projectcalico/calico/releases/download/v3.26.1/tigera-operator-v3.26.1.tgz
查看这个chart的中可定制的配置:
1helm show values tigera-operator-v3.26.1.tgz
2
3imagePullSecrets: {}
4
5installation:
6 enabled: true
7 kubernetesProvider: ''
8
9apiServer:
10 enabled: true
11
12certs:
13 node:
14 key:
15 cert:
16 commonName:
17 typha:
18 key:
19 cert:
20 commonName:
21 caBundle:
22
23# Resource requests and limits for the tigera/operator pod.
24resources: {}
25
26# Tolerations for the tigera/operator pod.
27tolerations:
28- effect: NoExecute
29 operator: Exists
30- effect: NoSchedule
31 operator: Exists
32
33# NodeSelector for the tigera/operator pod.
34nodeSelector:
35 kubernetes.io/os: linux
36
37# Custom annotations for the tigera/operator pod.
38podAnnotations: {}
39
40# Custom labels for the tigera/operator pod.
41podLabels: {}
42
43# Image and registry configuration for the tigera/operator pod.
44tigeraOperator:
45 image: tigera/operator
46 version: v1.30.4
47 registry: quay.io
48calicoctl:
49 image: docker.io/calico/ctl
50 tag: v3.26.1
定制的values.yaml如下:
1# 可针对上面的配置进行定制,例如calico的镜像改成从私有库拉取。
2# 这里只是个人本地环境测试k8s新版本，因此只有下面几行配置
3apiServer:
4 enabled: false
5installation:
6 kubeletVolumePluginPath: None
使用helm安装calico：
1helm install calico tigera-operator-v3.26.1.tgz -n kube-system --create-namespace -f values.yaml
等待并确认所有pod处于Running状态:
1kubectl get pod -n kube-system | grep tigera-operator
2tigera-operator-5fb55776df-wxbph 1/1 Running 0 5m10s
3
4kubectl get pods -n calico-system
5NAME READY STATUS RESTARTS AGE
6calico-kube-controllers-68884f975d-5d7p9 1/1 Running 0 5m24s
7calico-node-twbdh 1/1 Running 0 5m24s
8calico-typha-7b4bdd99c5-ssdn2 1/1 Running 0 5m24s
查看一下calico向k8s中添加的api资源:
1kubectl api-resources | grep calico
2bgpconfigurations crd.projectcalico.org/v1 false BGPConfiguration
3bgpfilters crd.projectcalico.org/v1 false BGPFilter
4bgppeers crd.projectcalico.org/v1 false BGPPeer
5blockaffinities crd.projectcalico.org/v1 false BlockAffinity
6caliconodestatuses crd.projectcalico.org/v1 false CalicoNodeStatus
7clusterinformations crd.projectcalico.org/v1 false ClusterInformation
8felixconfigurations crd.projectcalico.org/v1 false FelixConfiguration
9globalnetworkpolicies crd.projectcalico.org/v1 false GlobalNetworkPolicy
10globalnetworksets crd.projectcalico.org/v1 false GlobalNetworkSet
11hostendpoints crd.projectcalico.org/v1 false HostEndpoint
12ipamblocks crd.projectcalico.org/v1 false IPAMBlock
13ipamconfigs crd.projectcalico.org/v1 false IPAMConfig
14ipamhandles crd.projectcalico.org/v1 false IPAMHandle
15ippools crd.projectcalico.org/v1 false IPPool
16ipreservations crd.projectcalico.org/v1 false IPReservation
17kubecontrollersconfigurations crd.projectcalico.org/v1 false KubeControllersConfiguration
18networkpolicies crd.projectcalico.org/v1 true NetworkPolicy
19networksets crd.projectcalico.org/v1 true NetworkSet
这些api资源是属于calico的，因此不建议使用kubectl来管理，推荐按照calicoctl来管理这些api资源。 将calicoctl安装为kubectl的插件:
1cd /usr/local/bin
2curl -o kubectl-calico -O -L "https://github.com/projectcalico/calicoctl/releases/download/v3.21.5/calicoctl-linux-amd64"
3chmod +x kubectl-calico
验证插件正常工作:
1kubectl calico -h
2.5 验证k8s DNS是否可用 #
1kubectl run curl --image=radial/busyboxplus:curl -it
2If you don't see a command prompt, try pressing enter.
3[ root@curl:/ ]$
进入后执行nslookup kubernetes.default确认解析正常:
1nslookup kubernetes.default
2Server: 10.96.0.10
3Address 1: 10.96.0.10 kube-dns.kube-system.svc.cluster.local
4
5Name: kubernetes.default
6Address 1: 10.96.0.1 kubernetes.default.svc.cluster.local
2.6 向Kubernetes集群中添加Node节点 #
下面将node5, node6添加到Kubernetes集群中，分别在node5, node6上执行:
1kubeadm join 192.168.96.154:6443 --token ag6egz.xjq1zz01meq8iboq \
2 --discovery-token-ca-cert-hash sha256:3a13ba07a146b904a10fe2e3f0ea0056890f522c327eed073f8952a5b182883a
node5和node6加入集群时遇到了如下问题，调度到node5(openEuler 22.03系统)或node6(Rock Linux 8.8)的calico-typha Pod无法启动，并报下面的错误:
1kubectl describe po calico-typha-7bb687bdb4-zptbt -n calico-system
2...
3kubelet Failed to create pod sandbox: open /run/systemd/resolve/resolv.conf: no such file or directory
而/run/systemd/resolve/resolv.conf 文件是由systemd-resolved服务管理的。Ubuntu 22.04上默认安装并启动了这个服务。openEuler 22.03上没有安装这个服务。Rocky Linux 8.8上默认安装但没有启动这个服务。
下面node5上安装并启动systemd-resolved:
1yum install -y systemd-resolved
2systemctl enable systemd-resolved --now
3systemctl status systemd-resolved
在node6上启动systemd-resolved:
1systemctl enable systemd-resolved --now
2systemctl status systemd-resolved
之后3个节点上的calico相关pod全部启动正常:
1kubectl get po -n calico-system -o wide
2NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
3calico-kube-controllers-58b9468dd6-g26fb 1/1 Running 0 48m 10.244.3.66 node4 <none> <none>
4calico-node-h9m7k 1/1 Running 0 37m 192.168.96.156 node6 <none> <none>
5calico-node-rdv64 1/1 Running 0 48m 192.168.96.154 node4 <none> <none>
6calico-node-z5c4q 1/1 Running 0 37m 192.168.96.155 node5 <none> <none>
7calico-typha-7bb687bdb4-52kcv 1/1 Running 0 48m 192.168.96.154 node4 <none> <none>
8calico-typha-7bb687bdb4-zptbt 1/1 Running 0 23m 192.168.96.156 node6 <none> <none>
在master节点上执行命令查看集群中的节点(需要等待新加入节点上的calico-node pod启动正常)：
1kubectl get node
2NAME STATUS ROLES AGE VERSION
3node4 Ready control-plane 62m v1.28.0
4node5 Ready <none> 39m v1.28.0
5node6 Ready <none> 39m v1.28.0
3.Kubernetes常用组件部署 #
3.1 使用Helm部署ingress-nginx #
为了便于将集群中的服务暴露到集群外部，需要使用Ingress。接下来使用Helm将ingress-nginx部署到Kubernetes上。 Nginx Ingress Controller被部署在Kubernetes的边缘节点上。
这里将node4(192.168.96.154)作为边缘节点，打上Label：
1kubectl label node node4 node-role.kubernetes.io/edge=
下载ingress-nginx的helm chart:
1wget https://github.com/kubernetes/ingress-nginx/releases/download/helm-chart-4.7.0/ingress-nginx-4.7.0.tgz
查看ingress-nginx-4.7.0.tgz这个chart的可定制配置:
1helm show values ingress-nginx-4.7.0.tgz
对values.yaml配置定制如下:
1controller:
2 ingressClassResource:
3 name: nginx
4 enabled: true
5 default: true
6 controllerValue: "k8s.io/ingress-nginx"
7 admissionWebhooks:
8 enabled: false
9 replicaCount: 1
10 image:
11 # registry: registry.k8s.io
12 # image: ingress-nginx/controller
13 # tag: "v1.8.0"
14 registry: docker.io
15 image: unreachableg/registry.k8s.io_ingress-nginx_controller
16 tag: "v1.8.0"
17 digest: sha256:626fc8847e967dc06049c0eda9e093d77a08feff80179ae97538ba8b118570f3
18 hostNetwork: true
19 nodeSelector:
20 node-role.kubernetes.io/edge: ''
21 affinity:
22 podAntiAffinity:
23 requiredDuringSchedulingIgnoredDuringExecution:
24 - labelSelector:
25 matchExpressions:
26 - key: app
27 operator: In
28 values:
29 - nginx-ingress
30 - key: component
31 operator: In
32 values:
33 - controller
34 topologyKey: kubernetes.io/hostname
35 tolerations:
36 - key: node-role.kubernetes.io/master
37 operator: Exists
38 effect: NoSchedule
39 - key: node-role.kubernetes.io/master
40 operator: Exists
41 effect: PreferNoSchedule
nginx ingress controller的副本数replicaCount为1，将被调度到node4这个边缘节点上。这里并没有指定nginx ingress controller service的externalIPs，而是通过hostNetwork: true设置nginx ingress controller使用宿主机网络。
因为registry.k8s.io被墙，这里替换成unreachableg/registry.k8s.io_ingress-nginx_controller提前拉取一下镜像:
1crictl --runtime-endpoint=unix:///run/containerd/containerd.sock pull unreachableg/registry.k8s.io_ingress-nginx_controller:v1.8.0
1helm install ingress-nginx ingress-nginx-4.7.0.tgz --create-namespace -n ingress-nginx -f values.yaml
1kubectl get po -n ingress-nginx
2NAME READY STATUS RESTARTS AGE
3ingress-nginx-controller-86878885cd-m9xc4 1/1 Running 0 45s
测试访问http://192.168.96.154返回默认的nginx 404页，则部署完成。
3.2 使用Helm部署dashboard #
先部署metrics-server：
1wget https://github.com/kubernetes-sigs/metrics-server/releases/download/v0.6.4/components.yaml
修改components.yaml中的image为docker.io/unreachableg/k8s.gcr.io_metrics-server_metrics-server:v0.6.4。
修改components.yaml中容器的启动参数，加入--kubelet-insecure-tls。
1kubectl apply -f components.yaml
metrics-server的pod正常启动后，等一段时间就可以使用kubectl top查看集群和pod的metrics信息:
1kubectl top node
2NAME CPU(cores) CPU% MEMORY(bytes) MEMORY%
3node4 246m 6% 2296Mi 29%
4node5 145m 3% 810Mi 10%
5node6 97m 2% 933Mi 12%
6
7kubectl top pod -n kube-system
8NAME CPU(cores) MEMORY(bytes)
9coredns-66f779496c-7mlsm 3m 12Mi
10coredns-66f779496c-9m4cv 3m 13Mi
11etcd-node4 36m 44Mi
12kube-apiserver-node4 147m 353Mi
13kube-controller-manager-node4 24m 49Mi
14kube-proxy-bt64z 32m 26Mi
15kube-proxy-k4sft 94m 25Mi
16kube-proxy-x28q9 49m 17Mi
17kube-scheduler-node4 9m 18Mi
18metrics-server-7d686f4d9d-pgk6c 6m 17Mi
19tigera-operator-94d7f7696-nl5l7 3m 25Mi
接下来使用helm部署k8s的dashboard。当前k8s dashboard已经更新到了v3.0.0-alpha0，这里体验一下v3版本。
从k8s dashboard的v3版本开始，底层架构已更改，需要进行干净的安装，如果是在做升级dashboard操作，请首先移除先前的安装，这里是全新安装可以忽略。
k8s dashboard的v3版本现在默认使用cert-manager和nginx-ingress-controller。如果选择基于yaml清单的安装，请确保在集群中已安装它们。
我们前面已经安装了nginx-ingress-controller，下面先安装cert-manager:
1wget https://github.com/cert-manager/cert-manager/releases/download/v1.12.3/cert-manager.yaml
2
3kubectl apply -f cert-manager.yaml
确保cert-manager的所有pod启动正常:
1kubectl get po -n cert-manager
2NAME READY STATUS RESTARTS AGE
3cert-manager-6774cd657f-q9qpf 1/1 Running 0 102s
4cert-manager-cainjector-55c8b7b49b-vf8r4 1/1 Running 0 102s
5cert-manager-webhook-57797c469d-cgw4n 1/1 Running 0 102s
下载dashboard的yaml清单文件:
1wget https://raw.githubusercontent.com/kubernetes/dashboard/v3.0.0-alpha0/charts/kubernetes-dashboard.yaml
编辑kubernetes-dashboard.yaml清单文件，将其中的ingress中的host替换你的域名:
1kind: Ingress
2apiVersion: networking.k8s.io/v1
3metadata:
4 name: kubernetes-dashboard
5 namespace: kubernetes-dashboard
6 labels:
7 app.kubernetes.io/name: nginx-ingress
8 app.kubernetes.io/part-of: kubernetes-dashboard
9 annotations:
10 nginx.ingress.kubernetes.io/ssl-redirect: "true"
11 cert-manager.io/issuer: selfsigned
12spec:
13 ingressClassName: nginx
14 tls:
15 - hosts:
16 - localhost
17 secretName: kubernetes-dashboard-certs
18 rules:
19 - host: k8s.example.com
20 http:
21 paths:
22 - path: /
23 pathType: Prefix
24 backend:
25 service:
26 name: kubernetes-dashboard-web
27 port:
28 name: web
29 - path: /api
30 pathType: Prefix
31 backend:
32 service:
33 name: kubernetes-dashboard-api
34 port:
35 name: api
安装dashboard的yaml清单文件:
1kubectl apply -f kubernetes-dashboard.yaml
确认dashboard的相关Pod启动正常:
1kubectl get po -n kubernetes-dashboard
2NAME READY STATUS RESTARTS AGE
3kubernetes-dashboard-api-8586787f7-vtszr 1/1 Running 0 60s
4kubernetes-dashboard-metrics-scraper-6959b784dc-c98tz 1/1 Running 0 59s
5kubernetes-dashboard-web-6b6d549b4-qsrsn 1/1 Running 0 60s
6
7kubectl get ingress -n kubernetes-dashboard
8NAME CLASS HOSTS ADDRESS PORTS AGE
9kubernetes-dashboard nginx k8s.example.com 80, 443 6m47s
创建管理员sa:
1kubectl create serviceaccount kube-dashboard-admin-sa -n kube-system
2
3kubectl create clusterrolebinding kube-dashboard-admin-sa \
4--clusterrole=cluster-admin --serviceaccount=kube-system:kube-dashboard-admin-sa
创建集群管理员登录dashboard所需token:
1kubectl create token kube-dashboard-admin-sa -n kube-system --duration=87600h
2
3eyJhbGciOiJSUzI1NiIsImtpZCI6IlU1SlpSTS1YekNuVzE0T1k5TUdTOFFqN25URWxKckt6OUJBT0xzblBsTncifQ.eyJhdWQiOlsiaHR0cHM6Ly9rdWJlcm5ldGVzLmRlZmF1bHQuc3ZjLmNsdXN0ZXIubG9jYWwiXSwiZXhwIjoxOTY4OTA4MjgyLCJpYXQiOjE2NTM1NDgyODIsImlzcyI6Imh0dHBzOi8va3ViZXJuZXRlcy5kZWZhdWx0LnN2Yy5jbHVzdGVyLmxvY2FsIiwia3ViZXJuZXRlcy5pbyI6eyJuYW1lc3BhY2UiOiJrdWJlLXN5c3RlbSIsInNlcnZpY2VhY2NvdW50Ijp7Im5hbWUiOiJrdWJlLWRhc2hib2FyZC1hZG1pbi1zYSIsInVpZCI6IjY0MmMwMmExLWY1YzktNDFjNy04Mjc5LWQ1ZmI3MGRjYTQ3ZSJ9fSwibmJmIjoxNjUzNTQ4MjgyLCJzdWIiOiJzeXN0ZW06c2VydmljZWFjY291bnQ6a3ViZS1zeXN0ZW06a3ViZS1kYXNoYm9hcmQtYWRtaW4tc2EifQ.Xqxlo2vJ9Hb6UUVIqwvc8I5bahdxKzSRSaQI_67Yt7_YEHmkkHApxUGlwJYTKF9ufww3btlCmM8PtRn5_Q1yv-HAFyTOYKo8WHZ9UCm1bT3X8V8g4GQwZIl2dwmlUmKb1unBz2-em2uThQ015bMPDE8a42DV_bOwWjljVXat0nwV14nGorC8vKLjXbohrIJ3G1pgCJvlBn99F1RelmSUSQLlolUFoxpN6MamYTElwR6FfD-AGmFXvZSbcFaqVW0oxJHV70Gjs2igOtpqHFxxPlHT8aQzlRiybPtFyBf9Ll87TmVJimT89z8wv2si2Nee8bB2jhsApLn8TJyUSlbTXA
使用上面的token登录k8s dashboard。
