作者:运维有术
主机名 | IP | CPU | 内存 | 系统盘 | 数据盘 | 用途 |
---|---|---|---|---|---|---|
ksp-master-1 | 192.168.9.91 | 8 | 16 | 40 | 100 | 离线环境 KubeSphere/k8s-master |
ksp-master-2 | 192.168.9.92 | 8 | 16 | 40 | 100 | 离线环境 KubeSphere/k8s-master |
ksp-master-3 | 192.168.9.93 | 8 | 16 | 40 | 100 | 离线环境 KubeSphere/k8s-master |
ksp-registry | 192.168.9.90 | 4 | 8 | 40 | 100 | 离线环境镜像仓库节点(Harbor) |
ksp-deploy | 192.168.9.89 | 4 | 8 | 40 | 100 | 联网主机用于制作离线包 |
合计 | 4 | 32 | 64 | 200 | 500 |
KubeKey 从 v2.1.0 版开始新增了清单 (manifest) 和制品 (artifact) 的概念,为用户离线部署 KubeSphere 和 K8s 集群提供了一种简单便捷的解决方案。
manifest 是一个描述当前 Kubernetes 集群信息和定义 artifact 制品中需要包含哪些内容的文本文件。
使用 KubeKey,用户只需使用清单 manifest 文件来定义将要离线部署的集群环境需要的内容,再通过该 manifest 来导出制品 artifact 文件即可完成准备工作。离线部署时只需要 KubeKey 和 artifact 就可快速、简单的在环境中部署镜像仓库 Harbor 和 KubeSphere 以及 K8s 集群。
KubeKey 生成 manifest 文件有两种方式:
第一种方式的好处是可以根据 1:1 的运行集群构建离线集群,依赖于已有集群,灵活度不够,并不是所有人都具备这种条件。
因此,本文参考官方的离线文档,采用手写 manifest 清单文件的方式,实现离线环境的安装部署。
制作离线部署资源需要找一台能联通互联网的节点,本文为了资源的制作和离线部署验证,单独增加了一个能联网的 ksp-deploy 节点。
在该节点下载 KubeKey (下文简称 KK)最新版(v3.0.13)。具体 KK 版本号可以在 KubeKey 发行页面 查看。
cd ~
mkdir kubekey
cd kubekey/
# 选择中文区下载(访问 GitHub 受限时使用)
export KKZONE=cn
# 执行下载命令,获取最新版的 kk(受限于网络,有时需要执行多次)
curl -sfL https://get-kk.kubesphere.io | sh -
# 也可以使用下面的命令指定具体版本
curl -sfL https://get-kk.kubesphere.io | VERSION=v3.0.13 sh -
manifest 文件的编写可以参考 官方示例文档。有两个可用参考用例,一个简单版,一个完整版。参考简单版即可。
受限于篇幅,本文不展示原始的示例文件,建议读者仔细阅读官方示例,理解每一项配置的含义后根据需求改写(暂时无法理解的,可以直接使用下文提供的成品配置文件)。
执行下面的命令获取官方 releases v3.4.1 对应的 images-list(最终实验结果,一些镜像需要自行整理,完整的镜像列表可参考下文中的 manifest 文件)。
wget https://github.com/kubesphere/ks-installer/releases/download/v3.4.1/images-list.txt
完整的 Image(136 个) 分类及可裁剪性(必须留的有标粗,个人判断,未必精准):
为了保持完整性,本文使用了所有 Image,只是修改了镜像前缀为 registry.cn-beijing.aliyuncs.com/kubesphereio,修改后的完整的镜像列表在下面的 manifest 文件中展示,读者可根据需求裁剪。
本实验环境使用的操作系统是 x64 的 CentOS 7.9,所以只下载 centos7 的操作系统依赖包,其他操作系统请读者在 KubeKey releases 页面下载。
执行下面的命令,在能联网的部署服务器上执行下载。网络访问受限时,也可以通过其他方式,将该 ISO 下载后放到制作离线镜像的服务器的 /root/kubekey 目录下。
wget https://github.com/kubesphere/kubekey/releases/download/v3.0.12/centos7-rpms-amd64.iso
说明:KubeKey v3.0.13 的 release 中没包,只能在 v3.0.12 的 releases 中下载。
最终的 ISO(centos7-rpms-amd64.iso,314 MB)实际信息如下:
# 查看文件大小
[root@ksp-deploy kubekey]# ll -h centos7-rpms-amd64.iso
-rw-r--r--. 1 root root 315M Oct 23 18:21 centos7-rpms-amd64.iso
# 验证 sha256sum,确保 ISO 在下载过程中没出问题(官方提供的 sha256sum 信息在 https://github.com/kubesphere/kubekey/releases/download/v3.0.12/centos7-rpms.iso.sha256sum.txt)
[root@ksp-deploy kubekey]# sha256sum centos7-rpms-amd64.iso
2588fbc12acc9f3b95766a0c20382988f2a21da2a36e444b7e1a0f523e75f858 centos7-rpms-amd64.iso
根据上面的文件及相关信息,生成最终 manifest.yaml。
命名为 ksp-v3.4.1-manifest.yaml
apiVersion: kubekey.kubesphere.io/v1alpha2
kind: Manifest
metadata:
name: sample
spec:
arches:
- amd64
operatingSystems:
- arch: amd64
type: linux
id: centos
version: "7"
osImage: CentOS Linux 7 (Core)
repository:
iso:
localPath: "/root/kubekey/centos7-rpms-amd64.iso"
url:
kubernetesDistributions:
- type: kubernetes
version: v1.26.5
components:
helm:
version: v3.9.0
cni:
version: v1.2.0
etcd:
version: v3.4.13
calicoctl:
version: v3.26.1
containerRuntimes:
- type: docker
version: 20.10.23
- type: containerd
version: 1.6.4
crictl:
version: v1.24.0
docker-registry:
version: "2"
harbor:
version: v2.5.3
docker-compose:
version: v2.2.2
images:
- registry.cn-beijing.aliyuncs.com/kubesphereio/ks-installer:v3.4.1
- registry.cn-beijing.aliyuncs.com/kubesphereio/ks-apiserver:v3.4.1
- registry.cn-beijing.aliyuncs.com/kubesphereio/ks-console:v3.4.1
- registry.cn-beijing.aliyuncs.com/kubesphereio/ks-controller-manager:v3.4.1
- registry.cn-beijing.aliyuncs.com/kubesphereio/kubectl:v1.20.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/kubefed:v0.8.1
- registry.cn-beijing.aliyuncs.com/kubesphereio/tower:v0.2.1
- registry.cn-beijing.aliyuncs.com/kubesphereio/minio:RELEASE.2019-08-07T01-59-21Z
- registry.cn-beijing.aliyuncs.com/kubesphereio/mc:RELEASE.2019-08-07T23-14-43Z
- registry.cn-beijing.aliyuncs.com/kubesphereio/snapshot-controller:v4.0.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/nginx-ingress-controller:v1.3.1
- registry.cn-beijing.aliyuncs.com/kubesphereio/defaultbackend-amd64:1.4
- registry.cn-beijing.aliyuncs.com/kubesphereio/metrics-server:v0.4.2
- registry.cn-beijing.aliyuncs.com/kubesphereio/redis:5.0.14-alpine
- registry.cn-beijing.aliyuncs.com/kubesphereio/haproxy:2.0.25-alpine
- registry.cn-beijing.aliyuncs.com/kubesphereio/alpine:3.14
- registry.cn-beijing.aliyuncs.com/kubesphereio/openldap:1.3.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/netshoot:v1.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/cloudcore:v1.13.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/iptables-manager:v1.13.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/edgeservice:v0.3.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/gatekeeper:v3.5.2
- registry.cn-beijing.aliyuncs.com/kubesphereio/openpitrix-jobs:v3.3.2
- registry.cn-beijing.aliyuncs.com/kubesphereio/devops-apiserver:ks-v3.4.1
- registry.cn-beijing.aliyuncs.com/kubesphereio/devops-controller:ks-v3.4.1
- registry.cn-beijing.aliyuncs.com/kubesphereio/devops-tools:ks-v3.4.1
- registry.cn-beijing.aliyuncs.com/kubesphereio/ks-jenkins:v3.4.0-2.319.3-1
- registry.cn-beijing.aliyuncs.com/kubesphereio/inbound-agent:4.10-2
- registry.cn-beijing.aliyuncs.com/kubesphereio/builder-base:v3.2.2
- registry.cn-beijing.aliyuncs.com/kubesphereio/builder-nodejs:v3.2.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/builder-maven:v3.2.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/builder-maven:v3.2.1-jdk11
- registry.cn-beijing.aliyuncs.com/kubesphereio/builder-python:v3.2.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/builder-go:v3.2.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/builder-go:v3.2.2-1.16
- registry.cn-beijing.aliyuncs.com/kubesphereio/builder-go:v3.2.2-1.17
- registry.cn-beijing.aliyuncs.com/kubesphereio/builder-go:v3.2.2-1.18
- registry.cn-beijing.aliyuncs.com/kubesphereio/builder-base:v3.2.2-podman
- registry.cn-beijing.aliyuncs.com/kubesphereio/builder-nodejs:v3.2.0-podman
- registry.cn-beijing.aliyuncs.com/kubesphereio/builder-maven:v3.2.0-podman
- registry.cn-beijing.aliyuncs.com/kubesphereio/builder-maven:v3.2.1-jdk11-podman
- registry.cn-beijing.aliyuncs.com/kubesphereio/builder-python:v3.2.0-podman
- registry.cn-beijing.aliyuncs.com/kubesphereio/builder-go:v3.2.0-podman
- registry.cn-beijing.aliyuncs.com/kubesphereio/builder-go:v3.2.2-1.16-podman
- registry.cn-beijing.aliyuncs.com/kubesphereio/builder-go:v3.2.2-1.17-podman
- registry.cn-beijing.aliyuncs.com/kubesphereio/builder-go:v3.2.2-1.18-podman
- registry.cn-beijing.aliyuncs.com/kubesphereio/s2ioperator:v3.2.1
- registry.cn-beijing.aliyuncs.com/kubesphereio/s2irun:v3.2.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/s2i-binary:v3.2.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/tomcat85-java11-centos7:v3.2.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/tomcat85-java11-runtime:v3.2.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/tomcat85-java8-centos7:v3.2.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/tomcat85-java8-runtime:v3.2.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/java-11-centos7:v3.2.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/java-8-centos7:v3.2.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/java-8-runtime:v3.2.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/java-11-runtime:v3.2.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/nodejs-8-centos7:v3.2.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/nodejs-6-centos7:v3.2.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/nodejs-4-centos7:v3.2.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/python-36-centos7:v3.2.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/python-35-centos7:v3.2.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/python-34-centos7:v3.2.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/python-27-centos7:v3.2.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/argocd:v2.3.3
- registry.cn-beijing.aliyuncs.com/kubesphereio/argocd-applicationset:v0.4.1
- registry.cn-beijing.aliyuncs.com/kubesphereio/dex:v2.30.2
- registry.cn-beijing.aliyuncs.com/kubesphereio/redis:6.2.6-alpine
- registry.cn-beijing.aliyuncs.com/kubesphereio/configmap-reload:v0.7.1
- registry.cn-beijing.aliyuncs.com/kubesphereio/prometheus:v2.39.1
- registry.cn-beijing.aliyuncs.com/kubesphereio/prometheus-config-reloader:v0.55.1
- registry.cn-beijing.aliyuncs.com/kubesphereio/prometheus-operator:v0.55.1
- registry.cn-beijing.aliyuncs.com/kubesphereio/kube-rbac-proxy:v0.11.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/kube-state-metrics:v2.6.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/node-exporter:v1.3.1
- registry.cn-beijing.aliyuncs.com/kubesphereio/alertmanager:v0.23.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/thanos:v0.31.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/grafana:8.3.3
- registry.cn-beijing.aliyuncs.com/kubesphereio/kube-rbac-proxy:v0.11.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/notification-manager-operator:v2.3.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/notification-manager:v2.3.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/notification-tenant-sidecar:v3.2.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/elasticsearch-curator:v5.7.6
- registry.cn-beijing.aliyuncs.com/kubesphereio/opensearch-curator:v0.0.5
- registry.cn-beijing.aliyuncs.com/kubesphereio/elasticsearch-oss:6.8.22
- registry.cn-beijing.aliyuncs.com/kubesphereio/opensearch:2.6.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/opensearch-dashboards:2.6.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/fluentbit-operator:v0.14.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/docker:19.03
- registry.cn-beijing.aliyuncs.com/kubesphereio/fluent-bit:v1.9.4
- registry.cn-beijing.aliyuncs.com/kubesphereio/log-sidecar-injector:v1.2.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/filebeat:6.7.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/kube-events-operator:v0.6.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/kube-events-exporter:v0.6.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/kube-events-ruler:v0.6.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/kube-auditing-operator:v0.2.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/kube-auditing-webhook:v0.2.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/pilot:1.14.6
- registry.cn-beijing.aliyuncs.com/kubesphereio/proxyv2:1.14.6
- registry.cn-beijing.aliyuncs.com/kubesphereio/jaeger-operator:1.29
- registry.cn-beijing.aliyuncs.com/kubesphereio/jaeger-agent:1.29
- registry.cn-beijing.aliyuncs.com/kubesphereio/jaeger-collector:1.29
- registry.cn-beijing.aliyuncs.com/kubesphereio/jaeger-query:1.29
- registry.cn-beijing.aliyuncs.com/kubesphereio/jaeger-es-index-cleaner:1.29
- registry.cn-beijing.aliyuncs.com/kubesphereio/kiali-operator:v1.50.1
- registry.cn-beijing.aliyuncs.com/kubesphereio/kiali:v1.50
- registry.cn-beijing.aliyuncs.com/kubesphereio/busybox:1.31.1
- registry.cn-beijing.aliyuncs.com/kubesphereio/nginx:1.14-alpine
- registry.cn-beijing.aliyuncs.com/kubesphereio/wget:1.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/hello:plain-text
- registry.cn-beijing.aliyuncs.com/kubesphereio/wordpress:4.8-apache
- registry.cn-beijing.aliyuncs.com/kubesphereio/hpa-example:latest
- registry.cn-beijing.aliyuncs.com/kubesphereio/fluentd:v1.4.2-2.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/perl:latest
- registry.cn-beijing.aliyuncs.com/kubesphereio/examples-bookinfo-productpage-v1:1.16.2
- registry.cn-beijing.aliyuncs.com/kubesphereio/examples-bookinfo-reviews-v1:1.16.2
- registry.cn-beijing.aliyuncs.com/kubesphereio/examples-bookinfo-reviews-v2:1.16.2
- registry.cn-beijing.aliyuncs.com/kubesphereio/examples-bookinfo-details-v1:1.16.2
- registry.cn-beijing.aliyuncs.com/kubesphereio/examples-bookinfo-ratings-v1:1.16.3
- registry.cn-beijing.aliyuncs.com/kubesphereio/scope:1.13.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/pause:3.8
- registry.cn-beijing.aliyuncs.com/kubesphereio/pause:3.9
- registry.cn-beijing.aliyuncs.com/kubesphereio/kube-apiserver:v1.26.5
- registry.cn-beijing.aliyuncs.com/kubesphereio/kube-controller-manager:v1.26.5
- registry.cn-beijing.aliyuncs.com/kubesphereio/kube-scheduler:v1.26.5
- registry.cn-beijing.aliyuncs.com/kubesphereio/kube-proxy:v1.26.5
- registry.cn-beijing.aliyuncs.com/kubesphereio/k8s-dns-node-cache:1.15.12
- registry.cn-beijing.aliyuncs.com/kubesphereio/coredns:1.9.3
- registry.cn-beijing.aliyuncs.com/kubesphereio/kube-controllers:v3.26.1
- registry.cn-beijing.aliyuncs.com/kubesphereio/cni:v3.26.1
- registry.cn-beijing.aliyuncs.com/kubesphereio/node:v3.26.1
- registry.cn-beijing.aliyuncs.com/kubesphereio/pod2daemon-flexvol:v3.26.1
- registry.cn-beijing.aliyuncs.com/kubesphereio/haproxy:2.3
- registry.cn-beijing.aliyuncs.com/kubesphereio/provisioner-localpv:3.3.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/linux-utils:3.3.0
- registry.cn-beijing.aliyuncs.com/kubesphereio/kubectl:v1.22.0
kubectl:v1.22.0
registry:
auths: {}
manifest 修改说明
最后的 16 个镜像就是官方 images-list 文件缺失的镜像,一定要手工补充在 list 中
kubernetes 版本:v1.26.5
其他组件版本的选择:个人是根据 kubekey 在线安装过程的日志,查找相关组件的对应版本,以及参考官方的文档 组件默认版本说明、组件默认版本源码 、支持的组件列表 和 制品清单源码
开启 harbor 和 docker-compose 配置项,为后面通过 KubeKey 自建 harbor 仓库推送镜像使用。
默认创建的 manifest 里面的镜像列表从 docker.io 获取,替换前缀为 registry.cn-beijing.aliyuncs.com/kubesphereio。
若需要导出的 artifact 文件中包含操作系统依赖文件(如:conntarck、chrony 等),可在 operationSystem 元素中的 .repostiory.iso.url 中配置相应的 ISO 依赖文件下载地址为 localPath ,填写提前下载好的 ISO 包在本地的存放路径,并将 url 配置项置空。
您可以访问
https://github.com/kubesphere/kubekey/releases/tag/v3.0.12
下载 ISO 文件。
制品 (artifact) 说明
制品(artifact)是一个根据指定的 manifest 文件内容导出的包含镜像 tar 包和相关二进制文件的 tgz 包。
在 KubeKey 初始化镜像仓库、创建集群、添加节点和升级集群的命令中均可指定一个 artifact,KubeKey 将自动解包该 artifact 并在执行命令时直接使用解包出来的文件。
导出时请确保网络连接正常。
KubeKey 会解析镜像列表中的镜像名,若镜像名中的镜像仓库需要鉴权信息,可在 manifest 文件中的 .registry.auths 字段中进行配置(本文未配置)。
根据生成的 manifest,执行下面的命令制作制品(artifact)。
export KKZONE=cn
./kk artifact export -m ksp-v3.4.1-manifest.yaml -o ksp-v3.4.1-artifact.tar.gz
制作完成后,我们简单查看一下制作的制品包含哪些内容。
# 制作完成后查看结果
[root@ksp-deploy kubekey]# ls -lh
total 14G
-rw-r--r--. 1 root root 315M Oct 23 18:21 centos7-rpms-amd64.iso
-rwxr-xr-x. 1 root root 76M Nov 7 16:43 kk
-rw-r--r--. 1 root root 13G Dec 12 13:08 ksp-v3.4.1-artifact.tar.gz
-rw-r--r--. 1 root root 11K Dec 12 11:23 ksp-v3.4.1-manifest.yaml
drwxr-xr-x. 3 root root 18 Dec 12 13:08 kubekey
-rw-r--r--. 1 root root 35M Dec 8 10:33 kubekey-v3.0.13-linux-amd64.tar.gz
# 制作过程中捕捉的内容,可能不全(制作完成后 kubekey/artifact 目录会被清理)
[root@ksp-deploy kubekey]# du -sh kubekey/*
13G kubekey/artifact
88K kubekey/logs
[root@ksp-deploy kubekey]# du -sh kubekey/artifact/*
102M kubekey/artifact/cni
43M kubekey/artifact/containerd
14M kubekey/artifact/crictl
130M kubekey/artifact/docker
17M kubekey/artifact/etcd
45M kubekey/artifact/helm
12G kubekey/artifact/images
207M kubekey/artifact/kube
660M kubekey/artifact/registry
315M kubekey/artifact/repository
9.0M kubekey/artifact/runc
[root@ksp-deploy kubekey]# ls -lh ksp-v3.4.1-artifact.tar.gz
-rw-r--r--. 1 root root 13G Dec 12 13:08 ksp-v3.4.1-artifact.tar.gz
把 KubeKey 工具也制作成压缩包,便于拷贝到离线节点。
# 压缩
[root@ksp-deploy kubekey]# tar zcvf kubekey-v3.0.13.tar.gz kk kubekey-v3.0.13-linux-amd64.tar.gz
CentOS 7.9 的默认内核比较老,建议升级操作系统内核(具体是否升级请根据实际环境自行决定)。
wget https://mirrors.tuna.tsinghua.edu.cn/elrepo/kernel/el7/x86_64/RPMS/kernel-lt-5.4.263-1.el7.elrepo.x86_64.rpm
wget https://mirrors.tuna.tsinghua.edu.cn/elrepo/kernel/el7/x86_64/RPMS/kernel-lt-tools-libs-5.4.263-1.el7.elrepo.x86_64.rpm
wget https://mirrors.tuna.tsinghua.edu.cn/elrepo/kernel/el7/x86_64/RPMS/kernel-lt-tools-5.4.263-1.el7.elrepo.x86_64.rpm
tar zcvf kernel-lt-5.4.263-1-upgrade.tar.gz kernel-lt-*
至此,我们已经准备了 3 个离线部署资源包:
本演示环境使用 KubeKey 自动配置,生产环境可以参考我以前发布的部署实战系列文档手工配置。
每台服务器新增一块数据盘 /dev/sdb,用于 Containerd 和 Kubernetes Pod 的持久化存储。
为了满足部分用户希望在生产上线后,磁盘容量不够时可以实现扩容,本文采用了 LVM 的方式配置磁盘(实际上,本人维护的生产环境,几乎不用 LVM)。
本小节的配置比较简单,因此,下面的内容为无废话实操版。
请注意,以下操作无特殊说明时需在所有新增节点上执行。本文只选取 Master-1 节点作为演示,并假定其余服务器都已按照相同的方式进行配置和设置。
pvcreate /dev/sdb
vgcreate data /dev/sdb
# 使用所有空间,VG 名字为 data,LV 名字为 lvdata
lvcreate -l 100%VG data -n lvdata
mkfs.xfs /dev/mapper/data-lvdata
mkdir /data
mount /dev/mapper/data-lvdata /data/
tail -1 /etc/mtab >> /etc/fstab
mkdir -p /data/openebs/local
mkdir -p /data/containerd
ln -s /data/containerd /var/lib/containerd
说明:KubeKey 不支持在部署的时候更改 Containerd 的数据目录,只能用这种变通的方式(也可以提前手工安装 Containerd,建议)。
升级 CentOS 7.9 操作系统内核(可选项,生产环境建议执行)。
请注意,以下操作无特殊说明时需在集群所有节点上执行。本文只选取 Master-1 节点作为演示,并假定其余服务器都已按照相同的方式进行配置和设置。
将内核升级包 kernel-lt-5.4.263-1-upgrade.tar.gz
上传到集群中的每个节点的 /root 目录。
cd /root
tar xvf kernel-lt-5.4.263-1-upgrade.tar.gz
yum install kernel-lt-5.4.263-1.el7.elrepo.x86_64.rpm
注意: 这里我们只安装内核,同时安装其他包,会有包 conflicts 的报错。
grubby --set-default "/boot/vmlinuz-5.4.263-1.el7.elrepo.x86_64"
reboot
[root@ksp-master-1 ~]# uname -r
5.4.263-1.el7.elrepo.x86_64
在使用新内核引导系统后,安装其他 Kernel 相关软件包,解决之前出现的依赖冲突问题。
# 卸载旧版本的 kernel-tools 相关软件包,保留 kernel-3.10.0-1160.71.1,避免新内核异常无法进入系统(具体版本号以实际为准)
yum remove kernel-tools-3.10.0-1160.71.1.el7.x86_64 kernel-tools-libs-3.10.0-1160.71.1.el7.x86_64
# 安装新版本的 kernel-tools 相关软件包
yum install kernel-lt-tools-5.4.263-1.el7.elrepo.x86_64.rpm kernel-lt-tools-libs-5.4.263-1.el7.elrepo.x86_64.rpm
rm -rf kernel-lt-5.4.263-1.el7.elrepo.x86_64.rpm kernel-lt-tools-5.4.263-1.el7.elrepo.x86_64.rpm kernel-lt-tools-libs-5.4.263-1.el7.elrepo.x86_64.rpm
将以下离线部署资源包(KubeKey 和制品 artifact ),上传至离线环境部署节点 (通常是 Master-1 节点) 的 /data/ 目录(可根据实际情况修改)。
执行以下命令,解压 KubeKey:
# 创离线资源存放的数据目录
mkdir /data/kubekey
mv /data/ksp-v3.4.1-artifact.tar.gz /data/kubekey/
tar xvf /data/kubekey-v3.0.13.tar.gz -C /data/kubekey
cd /data/kubekey
./kk create config --with-kubesphere v3.4.1 --with-kubernetes v1.26.5 -f ksp-v341-v1265-offline.yaml
命令执行成功后,在当前目录会生成文件名为 ksp-v341-v1265-offline.yaml 的配置文件。
注意:生成的默认配置文件内容较多,这里就不做过多展示了,更多详细的配置参数请参考 官方配置示例。
在离线集群配置文件文件中 kind: Cluster 小节的作用是部署 Kubernetes 集群。本文示例采用 3 个节点同时作为 control-plane、etcd 节点和 worker 节点。
执行以下命令修改离线集群配置文件 ksp-v341-v1265-offline.yaml
:
vim ksp-v341-v1265-offline.yaml
修改 kind: Cluster 小节中 hosts 和 roleGroups 等信息,修改说明如下。
registry
作为仓库部署节点(本文为了满足读者的需求使用了 KubeKey 自动部署 Harbor 镜像仓库。当然,也可以使用已有的 Harbor,使用已有 Harbor 时此配置可以不加)type
类型为 harbor
,否则默认安装 docker registry修改后的完整示例如下:
apiVersion: kubekey.kubesphere.io/v1alpha2
kind: Cluster
metadata:
name: sample
spec:
hosts:
- {name: ksp-master-1, address: 192.168.9.91, internalAddress: 192.168.9.91, port:22, user: root, password: "P@88w0rd"}
- {name: ksp-master-2, address: 192.168.9.92, internalAddress: 192.168.9.92, user: root, password: "P@88w0rd"}
- {name: ksp-master-3, address: 192.168.9.93, internalAddress: 192.168.9.93, user: root, password: "P@88w0rd"}
- {name: ksp-registry, address: 192.168.9.90, internalAddress: 192.168.9.90, user: root, password: "P@88w0rd"}
roleGroups:
etcd:
- ksp-master-1
- ksp-master-2
- ksp-master-3
control-plane:
- ksp-master-1
- ksp-master-2
- ksp-master-3
worker:
- ksp-master-1
- ksp-master-2
- ksp-master-3
registry:
- ksp-registry
controlPlaneEndpoint:
## Internal loadbalancer for apiservers
internalLoadbalancer: haproxy
domain: lb.opsman.top
address: ""
port: 6443
kubernetes:
version: v1.26.5
clusterName: opsman.top
autoRenewCerts: true
containerManager: containerd
etcd:
type: kubekey
network:
plugin: calico
kubePodsCIDR: 10.233.64.0/18
kubeServiceCIDR: 10.233.0.0/18
## multus support. https://github.com/k8snetworkplumbingwg/multus-cni
multusCNI:
enabled: false
storage:
openebs:
basePath: /data/openebs/local # 默认没有的新增配置,base path of the local PV provisioner
registry:
# 如需使用 kk 部署 harbor, 可将该参数设置为 harbor,不设置该参数且需使用 kk 创建容器镜像仓库,将默认使用 docker registry。
type: harbor
# 注意:
# 1、如需使用 kk 部署的 harbor 或其他自定义仓库,可设置对应仓库的 auths,如使用 kk 创建 docker registry 仓库,则无需配置该参数。
# 2、kk 部署的 harbor,默认地址为 dockerhub.kubekey.local,如要修改确保与 privateRegistry 字段的值保持一致。
# 3、如需使用 kk 部署 harbor,请先注释该参数,待部署完 Harbor 后再来修改启用。
auths:
"registry.opsman.top":
username: admin
password: Harbor12345
certsPath: "/etc/docker/certs.d/registry.opsman.top"
# 设置集群部署时使用的私有仓库
privateRegistry: "registry.opsman.top"
namespaceOverride: "kubesphereio"
registryMirrors: []
insecureRegistries: []
addons: []
注意:离线 Harbor 部署,我们使用了私有自定义域名和自签名证书,实际生产中建议各位使用真正的互联网域名和正规机构签发的免费或收费证书。
在离线集群配置文件中 kind: ClusterConfiguration 小节的作用是部署 KubeSphere 及相关组件。
本文为了验证离线部署的完整性,启用了除 Kubeedge 、gatekeeper 以外的所有插件。
继续编辑离线集群配置文件 ksp-v341-v1265-offline.yaml
,修改 kind: ClusterConfiguration 部分来启用可插拔组件,具体的修改说明如下。
etcd:
monitoring: true # 将 "false" 更改为 "true"
endpointIps: localhost
port: 2379
tlsEnable: true
openpitrix:
store:
enabled: true # 将 "false" 更改为 "true"
devops:
enabled: true # 将 "false" 更改为 "true"
logging:
enabled: true # 将 "false" 更改为 "true"
events:
enabled: true # 将 "false" 更改为 "true"
alerting:
enabled: true # 将 "false" 更改为 "true"
auditing:
enabled: true # 将 "false" 更改为 "true"
servicemesh:
enabled: true # 将 "false" 更改为 "true"
istio:
components:
ingressGateways:
- name: istio-ingressgateway # 将服务暴露至服务网格之外。默认不开启。
enabled: false
cni:
enabled: false # 启用后,会在 Kubernetes pod 生命周期的网络设置阶段完成 Istio 网格的 pod 流量转发设置工作。
metrics_server:
enabled: true # 将 "false" 更改为 "true"
说明:KubeSphere 支持用于 部署 的容器组(Pod)弹性伸缩程序 (HPA)。在 KubeSphere 中,Metrics Server 控制着 HPA 是否启用。
network:
networkpolicy:
enabled: true # 将 "false" 更改为 "true"
ippool:
type: calico # 将 "none" 更改为 "calico"
topology:
type: weave-scope # 将 "none" 更改为 "weave-scope"
说明:
- 从 3.0.0 版本开始,用户可以在 KubeSphere 中配置原生 Kubernetes 的网络策略。
- 容器组 IP 池用于规划容器组网络地址空间,每个容器组 IP 池之间的地址空间不能重叠。
- 启用服务拓扑图以集成 Weave Scope (Docker 和 Kubernetes 的可视化和监控工具),服务拓扑图显示在您的项目中,将服务之间的连接关系可视化(实际上没啥用,该项目目前已停止维护了,姑且放着吧)。
spec:
namespace_override: kubesphereio
......
经过上述步骤,我们成功完成了对离线集群配置文件 ksp-v341-v1265-offline.yaml
的修改。由于篇幅限制,无法在此展示完整的文件内容,请各位读者根据上文提供的配置说明仔细核对。
为了验证 KubeKey 部署离线 Harbor 的能力,本实战内容采用 KubeKey 部署 Harbor。生产环境建议提前自建。
请注意,以下操作无特殊说明时需在离线环境部署节点(默认为 Master-1)上执行。
执行以下命令安装镜像仓库 Harbor(受限于篇幅,输出结果未展示):
cd /root/kubekey
./kk init registry -f ksp-v341-v1265-offline.yaml -a ksp-v3.4.1-artifact.tar.gz
说明:
ksp-v3.4.1-artifact.tar.gz
我们制作的离线部署资源包中的制品包。
部署完成后,我们 SSH 登陆到 Registry 节点,执行以下的命令,验证 Harbor 的安装情况:
[root@ksp-registry ~]# ls -lh /opt/harbor/
total 633M
......(受限于篇幅,输出结果未展示)
[root@ksp-registry ~]# docker images
REPOSITORY TAG IMAGE ID CREATED SIZE
goharbor/harbor-exporter v2.5.3 d9a8cfa37cf8 17 months ago 87.2MB
goharbor/chartmuseum-photon v2.5.3 788b207156ad 17 months ago 225MB
goharbor/redis-photon v2.5.3 5dc5331f3de8 17 months ago 154MB
goharbor/trivy-adapter-photon v2.5.3 27798821348a 17 months ago 251MB
goharbor/notary-server-photon v2.5.3 c686413b72ce 17 months ago 112MB
goharbor/notary-signer-photon v2.5.3 a3bc1def3f94 17 months ago 109MB
goharbor/harbor-registryctl v2.5.3 942de6829d43 17 months ago 136MB
goharbor/registry-photon v2.5.3 fb1278854b91 17 months ago 77.9MB
goharbor/nginx-photon v2.5.3 91877cbc147a 17 months ago 44.3MB
goharbor/harbor-log v2.5.3 ca36fb3b68a6 17 months ago 161MB
goharbor/harbor-jobservice v2.5.3 75e6a7496590 17 months ago 227MB
goharbor/harbor-core v2.5.3 93a775677473 17 months ago 203MB
goharbor/harbor-portal v2.5.3 d78f9bbad9ee 17 months ago 52.6MB
goharbor/harbor-db v2.5.3 bd50ae1eccdf 17 months ago 224MB
goharbor/prepare v2.5.3 15102b9ebde6 17 months ago 166MB
[root@ksp-registry ~]# cd /opt/harbor/
[root@ksp-registry harbor]# docker-compose ps -a
[root@ksp-registry harbor]# cat /opt/harbor/harbor.yml | grep hostname:
hostname: registry.opsman.top
[root@ksp-registry harbor]# ll /etc/docker/certs.d/registry.opsman.top/
total 12
-rw-r--r--. 1 root root 1103 Dec 13 09:47 ca.crt
-rw-r--r--. 1 root root 1253 Dec 13 09:47 registry.opsman.top.cert
-rw-------. 1 root root 1679 Dec 13 09:47 registry.opsman.top.key
小知识:KubeKey 部署 Harbor 时会自动同步自签名证书到集群所有节点,自己部署的 Harbor 必须手动复制 Registry 节点的自签名证书到集群所有节点。
KubeKey 完成复制的动作在下面两个模块:
[InitRegistryModule] Synchronize certs file
[InitRegistryModule] Synchronize certs file to all nodes
由于 Harbor 项目存在访问控制(RBAC)的限制,即只有指定角色的用户才能执行某些操作。如果您未创建项目,则镜像不能被推送到 Harbor。
Harbor 中有两种类型的项目:
使用 KubeKey 安装的 Harbor 默认信息如下:
接下来的任务需要我们 SSH 登陆到 Registry 节点执行。
执行以下命令,下载官方提供的自动初始化 Harbor 仓库的脚本(可不下载直接用本文提供的脚本):
curl -O https://raw.githubusercontent.com/kubesphere/ks-installer/master/scripts/create_project_harbor.sh
vim create_project_harbor.sh
。#!/usr/bin/env bash
# Harbor 仓库地址(写域名,默认配置为 https://dockerhub.kubekey.local)
url="https://registry.opsman.top"
# 访问 Harbor 仓库的默认用户和密码(生产环境建议修改)
user="admin"
passwd="Harbor12345"
# 需要创建的项目名列表,按我们制作的离线包的镜像命名规则,实际上只需要创建一个 kubesphereio 即可,这里保留了所有可用值,各位可根据自己的离线仓库镜像名称规则修改。
harbor_projects=(library
kubesphere
calico
coredns
openebs
csiplugin
minio
mirrorgooglecontainers
osixia
prom
thanosio
jimmidyson
grafana
elastic
istio
jaegertracing
jenkins
weaveworks
openpitrix
joosthofman
nginxdemos
fluent
kubeedge
kubesphereio
)
for project in "${harbor_projects[@]}"; do
echo "creating $project"
curl -k -u "${user}:${passwd}" -X POST -H "Content-Type: application/json" "${url}/api/v2.0/projects" -d "{ \"project_name\": \"${project}\", \"public\": true}"
done
重点注意:
- harbor_projects 中一定要新增
kubesphereio
,默认没有。不加后面报错,详情见 问题 2。- 脚本创建的是 public 项目,如需要私有项目请修改脚本。
sh create_project_harbor.sh
将提前准备好的离线镜像推送到 Harbor 仓库,这一步为可选项,因为创建集群的时候默认会推送镜像(本文使用参数忽略了)。为了部署成功率,建议先推送。
./kk artifact image push -f ksp-v341-v1265-offline.yaml -a ksp-v3.4.1-artifact.tar.gz
执行以下命令,安装 KubeSphere 和 K8s 集群。
./kk create cluster -f ksp-v341-v1265-offline.yaml -a ksp-v3.4.1-artifact.tar.gz --with-packages --skip-push-images
参数说明
- ksp-v341-v1265-offline.yaml:离线环境集群的配置文件
- ksp-v3.4.1-artifact.tar.gz:制品包的 tar 包镜像
- --with-packages:如需要安装操作系统依赖,需指定该选项
- --skip-push-images: 忽略推送镜像,因为,前面已经完成了推送镜像到私有仓库的任务
上面的命令执行后,首先 KubeKey 会检查部署 K8s 的依赖及其他详细要求。检查合格后,系统将提示您确认安装。输入 yes 并按 ENTER 继续部署。
特殊说明:由于本文在安装的过程中启用了日志插件,因此在安装的过程中必须按照 「问题 6」的描述手工介入处理,否则安装会失败。
安装过程日志输出比较多,受限于篇幅本文只展示最终结果。
部署完成需要大约 10-30 分钟左右,具体看网速、机器配置、启用多少插件等。
部署完成后,您应该会在终端上看到类似于下面的输出。提示部署完成的同时,输出中还会显示用户登陆 KubeSphere 的默认管理员用户和密码。
#####################################################
### Welcome to KubeSphere! ###
#####################################################
Console: http://192.168.9.91:30880
Account: admin
Password: P@88w0rd
NOTES:
1. After you log into the console, please check the
monitoring status of service components in
"Cluster Management". If any service is not
ready, please wait patiently until all components
are up and running.
2. Please change the default password after login.
#####################################################
https://kubesphere.io 2023-12-13 10:56:38
#####################################################
10:56:40 CST skipped: [ksp-master-3]
10:56:40 CST skipped: [ksp-master-2]
10:56:40 CST success: [ksp-master-1]
10:56:40 CST Pipeline[CreateClusterPipeline] execute successfully
Installation is complete.
Please check the result using the command:
kubectl logs -n kubesphere-system $(kubectl get pod -n kubesphere-system -l 'app in (ks-install, ks-installer)' -o jsonpath='{.items[0].metadata.name}') -f
小技巧:
kubectl get pod -A
或是 kubectl get pod -A | grep -v Running
观察进度,如出现异常可及时介入处理。kubectl logs -n kubesphere-system $(kubectl get pod -n kubesphere-system -l 'app in (ks-install, ks-installer)' -o jsonpath='{.items[0].metadata.name}') -f
登录 Web 控制台通过 http://{IP}:30880
使用默认帐户和密码 admin/P@88w0rd
访问 KubeSphere 的 Web 控制台,简单的验证一下部署结果。
执行脚本 create_project_harbor.sh
,创建 Harbor 项目时,报错如下:
curl performs SSL certificate verification by default, using a "bundle"
of Certificate Authority (CA) public keys (CA certs). If the default
bundle file isn't adequate, you can specify an alternate file
using the --cacert option.
If this HTTPS server uses a certificate signed by a CA represented in
the bundle, the certificate verification probably failed due to a
problem with the certificate (it might be expired, or the name might
not match the domain name in the URL).
If you'd like to turn off curl's verification of the certificate, use
the -k (or --insecure) option.
因为我们 Harbor 使用的是 Https 协议,并且使用了私有的域名和私有自定义的证书。需修改脚本,在 curl 命令中加入 -k
参数。
# 修改后命令
curl -k -u "${user}:${passwd}" -X POST -H "Content-Type: application/json" "${url}/api/v2.0/projects" -d "{ \"project_name\": \"${project}\", \"public\": true}"
部署 Kubesphere 和 Kubernetes 集群之前,执行推送离线镜像的命令,报错如下:
15:01:29 CST [CopyImagesToRegistryModule] Push multi-arch manifest to private registry
15:01:29 CST message: [LocalHost]
get manifest list failed by module cache
15:01:29 CST failed: [LocalHost]
error: Pipeline[ArtifactImagesPushPipeline] execute failed: Module[CopyImagesToRegistryModule] exec failed:
failed: [LocalHost] [PushManifest] exec failed after 1 retries: get manifest list failed by module cache
参考 issues-2054,在 Harbor 中创建项目 kubesphereio
。
部署集群时,报错如下:
17:26:33 CST message: [ksp-master-1]
pull image failed: Failed to exec command: sudo -E /bin/bash -c "env PATH=$PATH crictl pull registry.opsman.top/kubesphereio/pause:3.8 --platform amd64"
E1211 17:26:33.225514 5977 remote_image.go:238] "PullImage from image service failed" err="rpc error: code = Unknown desc = failed to pull and unpack image \"registry.opsman.top/kubesphereio/pause:3.8\": failed to resolve reference \"registry.opsman.top/kubesphereio/pause:3.8\": failed to do request: Head \"https://registry.opsman.top/v2/kubesphereio/pause/manifests/3.8\": x509: certificate signed by unknown authority" image="registry.opsman.top/kubesphereio/pause:3.8"
FATA[0000] pulling image: rpc error: code = Unknown desc = failed to pull and unpack image "registry.opsman.top/kubesphereio/pause:3.8": failed to resolve reference "registry.opsman.top/kubesphereio/pause:3.8": failed to do request: Head "https://registry.opsman.top/v2/kubesphereio/pause/manifests/3.8": x509: certificate signed by unknown authority: Process exited with status 1
17:26:33 CST retry: [ksp-master-1]
参考 issues-1762,在集群部署文件 ksp-v341-v1265-offline.yaml
的 registry 部分配置中加入以下内容(一定要确保 /etc/docker/certs.d/registry.opsman.top 文件夹存在):
# 本文用的 registry.opsman.top 默认为 dockerhub.kubekey.local
certsPath: "/etc/docker/certs.d/registry.opsman.top"
如果在部署过程中报错,可以在 containerd 的配置文件中,加入下面的配置,重启 containerd 服务,再重新执行部署任务。
# vi /etc/containerd/config.toml 找到正确位置加入(一般在文件最后有类似段落,修改替换即可)
[plugins."io.containerd.grpc.v1.cri".registry.configs."registry.opsman.top".tls]
ca_file = "/etc/docker/certs.d/registry.opsman.top/ca.crt"
cert_file = "/etc/docker/certs.d/registry.opsman.top/registry.opsman.top.cert"
key_file = "/etc/docker/certs.d/registry.opsman.top/registry.opsman.top.key"
insecure_skip_verify = false
# 重启服务
systemctl restart containerd
如果报错提示 /etc/docker/certs.d/registry.opsman.top
找不到,可以按下面的操作执行(没用 KubeKey 部署 Harbor 时会出现)。
# 1. 登陆到 registry 节点
# 2. 复制 /etc/docker/certs.d/registry.opsman.top 文件夹到其他节点
部署集群时,报错如下:
17:50:16 CST message: [ksp-master-3]
pull image failed: Failed to exec command: sudo -E /bin/bash -c "env PATH=$PATH crictl pull registry.opsman.top/kubesphereio/pause:3.8 --platform amd64"
E1211 17:50:16.841836 8464 remote_image.go:238] "PullImage from image service failed" err="rpc error: code = Unknown desc = failed to pull and unpack image \"registry.opsman.top/kubesphereio/pause:3.8\": failed to resolve reference \"registry.opsman.top/kubesphereio/pause:3.8\": get TLSConfig for registry \"https://registry.opsman.top\": failed to load cert file: open /etc/docker/certs.d/registry.opsman.top/registry.opsman.top.cert: no such file or directory" image="registry.opsman.top/kubesphereio/pause:3.8"
FATA[0000] pulling image: rpc error: code = Unknown desc = failed to pull and unpack image "registry.opsman.top/kubesphereio/pause:3.8": failed to resolve reference "registry.opsman.top/kubesphereio/pause:3.8": get TLSConfig for registry "https://registry.opsman.top": failed to load cert file: open /etc/docker/certs.d/registry.opsman.top/registry.opsman.top.cert: no such file or directory: Process exited with status 1
17:50:16 CST retry: [ksp-master-3]
制作离线资源的官方 image-list 中不存在 pause:3.8 这个 image,需要自己预先加上再制作制品。实际上一共需要补充 16 个,报错都不一样,可以根据报错信息补充(本文提供的 images-list 已经是补充完整的)。
Events:
Type Reason Age From Message
---- ------ ---- ---- -------
Normal Scheduled 18m default-scheduler Successfully assigned kube-system/metrics-server-5d65c798b8-m9tbj to ksp-master-3
Normal Pulling 16m (x4 over 18m) kubelet Pulling image "registry.opsman.top/kubesphere/metrics-server:v0.4.2"
Warning Failed 16m (x4 over 18m) kubelet Failed to pull image "registry.opsman.top/kubesphere/metrics-server:v0.4.2": rpc error: code = NotFound desc = failed to pull and unpack image "registry.opsman.top/kubesphere/metrics-server:v0.4.2": failed to resolve reference "registry.opsman.top/kubesphere/metrics-server:v0.4.2": registry.opsman.top/kubesphere/metrics-server:v0.4.2: not found
Warning Failed 16m (x4 over 18m) kubelet Error: ErrImagePull
Warning Failed 16m (x6 over 17m) kubelet Error: ImagePullBackOff
Normal BackOff 2m57s (x64 over 17m) kubelet Back-off pulling image "registry.opsman.top/kubesphere/metrics-server:v0.4.2"
可能是 bug(已反馈给社区,暂未确认)。社区给了临时方案在离线集群配置文件中 kind: ClusterConfiguration 小节中,加入参数 namespace_override: kubesphereio
。
# kubectl describe pod opensearch-cluster-data-0 -n kubesphere-logging-system
Events:
Type Reason Age From Message
---- ------ ---- ---- -------
Normal Pulling 56s (x3 over 2m57s) kubelet Pulling image "busybox:latest"
Warning Failed 15s (x3 over 2m17s) kubelet Error: ErrImagePull
Warning Failed 15s kubelet Failed to pull image "busybox:latest": rpc error: code = Unknown desc = failed to pull and unpack image "docker.io/library/busybox:latest": failed to resolve reference "docker.io/library/busybox:latest": failed to do request: Head "https://registry-1.docker.io/v2/library/busybox/manifests/latest": dial tcp: lookup registry-1.docker.io on 114.114.114.114:53: read udp 192.168.9.92:37639->114.114.114.114:53: i/o timeout
Normal BackOff 0s (x3 over 2m16s) kubelet Back-off pulling image "busybox:latest"
Warning Failed 0s (x3 over 2m16s) kubelet Error: ImagePullBackOff
这个应该也是个 bug,官方配置写死了 busybox 的地址(已反馈给社区,已确认),暂时需要手改。
# 查看 sts
[root@ksp-master-1 kubekey]# kubectl get sts -n kubesphere-logging-system
NAME READY AGE
opensearch-cluster-data 0/2 7m42s
opensearch-cluster-master 0/1 7m40s
# 修改 sts 使用的 busyboy 镜像为本地镜像(细节不展示)
kubectl edit sts opensearch-cluster-data -n kubesphere-logging-system
kubectl edit sts opensearch-cluster-master -n kubesphere-logging-system
# 本文修改后 image 内容(自己根据实际情况修改域名前缀)
registry.opsman.top/kubesphereio/busybox:1.31.1
[preflight] You can also perform this action in beforehand using 'kubeadm config images pull'
[WARNING ImagePull]: failed to pull image registry.opsman.top/kubesphereio/pause:3.9: output: E1213 10:31:38.814017 8325 remote_image.go:238] "PullImage from image service failed" err="rpc error: code = NotFound desc = failed to pull and unpack image \"registry.opsman.top/kubesphereio/pause:3.9\": failed to resolve reference \"registry.opsman.top/kubesphereio/pause:3.9\": registry.opsman.top/kubesphereio/pause:3.9: not found" image="registry.opsman.top/kubesphereio/pause:3.9"
time="2023-12-13T10:31:38+08:00" level=fatal msg="pulling image: rpc error: code = NotFound desc = failed to pull and unpack image \"registry.opsman.top/kubesphereio/pause:3.9\": failed to resolve reference \"registry.opsman.top/kubesphereio/pause:3.9\": registry.opsman.top/kubesphereio/pause:3.9: not found"
, error: exit status 1
这个报错是在部署过程中出现的,但是报错并未导致集群部署失败。解决办法跟问题 4
一样,制作离线资源的官方 image-list 中不存在 pause:3.9 这个 image,需要自己预先加上再制作制品。
Events:
Type Reason Age From Message
---- ------ ---- ---- -------
Normal Pulling 4m25s kubelet Pulling image "registry.opsman.top/kubesphereio/ks-apiserver:v3.4.1"
Warning Failed 4m22s kubelet Failed to pull image "registry.opsman.top/kubesphereio/ks-apiserver:v3.4.1": rpc error: code = Unknown desc = failed to pull and unpack image "registry.opsman.top/kubesphereio/ks-apiserver:v3.4.1": failed to extract layer sha256:9f4598b692bcf57921e45bc384b60591c5f8eac82e32e781e9fe5849ef6eb29e: write /var/lib/containerd/io.containerd.snapshotter.v1.overlayfs/snapshots/291/fs/usr/local/bin/ks-apiserver: no space left on device: unknown
Warning Failed 4m22s kubelet Error: ErrImagePull
Warning Failed 4m21s kubelet Error: ImagePullBackOff
Normal BackOff 39s (x15 over 4m21s) kubelet Back-off pulling image "registry.opsman.top/kubesphereio/ks-apiserver:v3.4.1"
人为造成,KubeKey 的离线部署资源包放在了节点 ksp-master-1
,同时,该节点也是 K8s 的 Master-1 节点。该节点的系统盘为 40G,演示环境没加额外的数据盘,所以磁盘空间满了。实际使用中,建议新加一块数据盘。
容器 devops-argocd-dex-server
反复重启。
# kubectl describe pod devops-argocd-dex-server-84fff59566-2brcn -n argocd
Events:
Type Reason Age From Message
---- ------ ---- ---- -------
Normal Created 35m (x2 over 36m) kubelet Created container dex-server
Normal Started 35m (x2 over 36m) kubelet Started container dex-server
Normal Killing 35m kubelet Container dex-server failed liveness probe, will be restarted
Normal Pulled 35m kubelet Container image "registry.opsman.top/kubesphereio/dex:v2.30.2" already present on machine
Warning Unhealthy 35m (x6 over 36m) kubelet Liveness probe failed: Get "http://10.233.89.12:5558/healthz/live": dial tcp 10.233.89.12:5558: connect: connection refused
Warning Unhealthy 22m (x38 over 36m) kubelet Readiness probe failed: Get "http://10.233.89.12:5558/healthz/ready": dial tcp 10.233.89.12:5558: connect: connection refused
Warning BackOff 2m21s (x127 over 33m) kubelet Back-off restarting failed container dex-server in pod devops-argocd-dex-server-84fff59566-2brcn_argocd(d83b98dc-cd21-47d9-aa5e-e7a26d458e7f)
# 细节没深究,删除 pod 自动重建,但是还会自动重启
kubectl delete pod devops-argocd-dex-server-84fff59566-2brcn -n argocd
本次的实战课程,我们详细介绍了如何使用 KubeKey v3.0.13 离线部署 KubeSphere 和 K8s 集群。
概括总结全文主要涉及以下内容:
**受限于篇幅限制,本文略有遗憾之处在于很多部署任务的输出结果没有展示,建议读者在实战中仔细分析。
免责声明:
本文由博客一文多发平台 OpenWrite 发布!