本指南用于指导如何将 kubeadm 集群从 1.7.x 版本升级到 1.8.x 版本，也可用于从 1.7.x 到 1.7.y 及从 1.8.x 到 1.8.y，其中 y > x。 如果您当前使用 1.6 版本的集群，请查看 将 kubeadm 集群从 1.6 升级到 1.7。

Before you begin

在开始前:

  • 您需要有一个正常工作的 1.7.0 或更高版本的 kubeadm Kubernetes 集群，以进行此处描述的流程。
  • 请确保已经仔细阅读 版本更新。
  • 由于 kubeadm upgrade 不会升级 etcd，请确保已对其进行了备份。例如，您可以使用 etcdctl backup 命令完成这个工作。
  • 请注意，kubeadm upgrade 只会升级 Kubernetes 内建（Kubernetes-internal）组件，不会触及任何工作负载。作为最佳实践，您应该备份所有重要数据。例如任何应用层级的状态数据，如应用可能依赖的数据库（如 MySQL 或 MongoDB）等，在开始升级前必须对其进行备份。
    此外，请注意升级仅支持一个小版本号。也就是说，您只能从 1.7 升级到 1.8 而不能从 1.7 升级到 1.9。

升级控制平面（control plane）

您需要在 master 节点上执行这些步骤：

  • 像这样使用 curl 安装最新版本的 kubeadm：
  1. $ export VERSION=$(curl -sSL https://dl.k8s.io/release/stable.txt) # or manually specify a released Kubernetes version
  2. $ export ARCH=amd64 # or: arm, arm64, ppc64le, s390x
  3. $ curl -sSL https://dl.k8s.io/release/${VERSION}/bin/linux/${ARCH}/kubeadm > /usr/bin/kubeadm
  4. $ chmod a+rx /usr/bin/kubeadm

警示：升级控制平面前在系统上升级 kubeadm 包将导致升级失败。即使 kubeadm 已经放入 Kubernetes 仓库中，您仍应该手动安装它。Kubeadm 团队正在修复这个限制。

验证下载的 kubeadm 是否工作正常，是否为预期的版本：

  1. $ kubeadm version
  • 如果这是您第一次使用 kubeadm upgrade，为了保存配置文件以便用于以后的升级，请执行：
    请注意，下列命令的运行需要您回顾首次运行 kubeadm init 时传递的参数。

如果您使用过标志，请执行：

  1. $ kubeadm config upload from-flags [flags]

这里的 flags 可以为空。

如果您使用的是配置文件，请执行：

  1. $ kubeadm config upload from-file --config [config]

这里的 config 是必须的。

  • 在 master 节点上执行下列步骤：
  1. $ kubeadm upgrade plan
  2. [preflight] Running pre-flight checks
  3. [upgrade] Making sure the cluster is healthy:
  4. [upgrade/health] Checking API Server health: Healthy
  5. [upgrade/health] Checking Node health: All Nodes are healthy
  6. [upgrade/health] Checking Static Pod manifests exists on disk: All manifests exist on disk
  7. [upgrade/config] Making sure the configuration is correct:
  8. [upgrade/config] Reading configuration from the cluster...
  9. [upgrade/config] FYI: You can look at this config file with 'kubectl -n kube-system get cm kubeadm-config -o yaml'
  10. [upgrade] Fetching available versions to upgrade to:
  11. [upgrade/versions] Cluster version: v1.7.1
  12. [upgrade/versions] kubeadm version: v1.8.0
  13. [upgrade/versions] Latest stable version: v1.8.0
  14. [upgrade/versions] Latest version in the v1.7 series: v1.7.6
  15.  
  16. Components that must be upgraded manually after you've upgraded the control plane with 'kubeadm upgrade apply':
  17. COMPONENT   CURRENT      AVAILABLE
  18. Kubelet     1 x v1.7.1   v1.7.6
  19.  
  20. Upgrade to the latest version in the v1.7 series:
  21.  
  22. COMPONENT            CURRENT   AVAILABLE
  23. API Server           v1.7.1    v1.7.6
  24. Controller Manager   v1.7.1    v1.7.6
  25. Scheduler            v1.7.1    v1.7.6
  26. Kube Proxy           v1.7.1    v1.7.6
  27. Kube DNS             1.14.4    1.14.4
  28.  
  29. You can now apply the upgrade by executing the following command:
  30.  
  31.     kubeadm upgrade apply v1.7.6
  32.  
  33. _____________________________________________________________________
  34.  
  35. Components that must be upgraded manually after you've upgraded the control plane with 'kubeadm upgrade apply':
  36. COMPONENT   CURRENT      AVAILABLE
  37. Kubelet     1 x v1.7.1   v1.8.0
  38.  
  39. Upgrade to the latest experimental version:
  40.  
  41. COMPONENT            CURRENT   AVAILABLE
  42. API Server           v1.7.1    v1.8.0
  43. Controller Manager   v1.7.1    v1.8.0
  44. Scheduler            v1.7.1    v1.8.0
  45. Kube Proxy           v1.7.1    v1.8.0
  46. Kube DNS             1.14.4    1.14.4
  47.  
  48. You can now apply the upgrade by executing the following command:
  49.  
  50.     kubeadm upgrade apply v1.8.0
  51.  
  52. Note: Before you do can perform this upgrade, you have to update kubeadm to v1.8.0
  53.  
  54. _____________________________________________________________________

kubeadm upgrade plan 将检查您的集群是否处于可升级状态，并以用户友好的方式获取可升级的版本。

  • 选择一个版本进行升级，例如执行下面的 kubeadm upgrade apply：
  1. $ kubeadm upgrade apply v1.8.0
  2. [preflight] Running pre-flight checks
  3. [upgrade] Making sure the cluster is healthy:
  4. [upgrade/health] Checking API Server health: Healthy
  5. [upgrade/health] Checking Node health: All Nodes are healthy
  6. [upgrade/health] Checking Static Pod manifests exists on disk: All manifests exist on disk
  7. [upgrade/config] Making sure the configuration is correct:
  8. [upgrade/config] Reading configuration from the cluster...
  9. [upgrade/config] FYI: You can look at this config file with 'kubectl -n kube-system get cm kubeadm-config -o yaml'
  10. [upgrade/version] You have chosen to upgrade to version "v1.8.0"
  11. [upgrade/versions] Cluster version: v1.7.1
  12. [upgrade/versions] kubeadm version: v1.8.0
  13. [upgrade/prepull] Will prepull images for components [kube-apiserver kube-controller-manager kube-scheduler]
  14. [upgrade/prepull] Prepulling image for component kube-scheduler.
  15. [upgrade/prepull] Prepulling image for component kube-apiserver.
  16. [upgrade/prepull] Prepulling image for component kube-controller-manager.
  17. [apiclient] Found 0 Pods for label selector k8s-app=upgrade-prepull-kube-scheduler
  18. [apiclient] Found 1 Pods for label selector k8s-app=upgrade-prepull-kube-scheduler
  19. [apiclient] Found 1 Pods for label selector k8s-app=upgrade-prepull-kube-apiserver
  20. [apiclient] Found 1 Pods for label selector k8s-app=upgrade-prepull-kube-controller-manager
  21. [upgrade/prepull] Prepulled image for component kube-apiserver.
  22. [upgrade/prepull] Prepulled image for component kube-controller-manager.
  23. [upgrade/prepull] Prepulled image for component kube-scheduler.
  24. [upgrade/prepull] Successfully prepulled the images for all the control plane components
  25. [upgrade/apply] Upgrading your Static Pod-hosted control plane to version "v1.8.0"...
  26. [upgrade/staticpods] Writing upgraded Static Pod manifests to "/etc/kubernetes/tmp/kubeadm-upgraded-manifests432902769"
  27. [controlplane] Wrote Static Pod manifest for component kube-apiserver to "/etc/kubernetes/tmp/kubeadm-upgraded-manifests432902769/kube-apiserver.yaml"
  28. [controlplane] Wrote Static Pod manifest for component kube-controller-manager to "/etc/kubernetes/tmp/kubeadm-upgraded-manifests432902769/kube-controller-manager.yaml"
  29. [controlplane] Wrote Static Pod manifest for component kube-scheduler to "/etc/kubernetes/tmp/kubeadm-upgraded-manifests432902769/kube-scheduler.yaml"
  30. [upgrade/staticpods] Moved upgraded manifest to "/etc/kubernetes/manifests/kube-apiserver.yaml" and backed up old manifest to "/etc/kubernetes/tmp/kubeadm-backup-manifests155856668/kube-apiserver.yaml"
  31. [upgrade/staticpods] Waiting for the kubelet to restart the component
  32. [apiclient] Found 1 Pods for label selector component=kube-apiserver
  33. [upgrade/staticpods] Component "kube-apiserver" upgraded successfully!
  34. [upgrade/staticpods] Moved upgraded manifest to "/etc/kubernetes/manifests/kube-controller-manager.yaml" and backed up old manifest to "/etc/kubernetes/tmp/kubeadm-backup-manifests155856668/kube-controller-manager.yaml"
  35. [upgrade/staticpods] Waiting for the kubelet to restart the component
  36. [apiclient] Found 1 Pods for label selector component=kube-controller-manager
  37. [upgrade/staticpods] Component "kube-controller-manager" upgraded successfully!
  38. [upgrade/staticpods] Moved upgraded manifest to "/etc/kubernetes/manifests/kube-scheduler.yaml" and backed up old manifest to "/etc/kubernetes/tmp/kubeadm-backup-manifests155856668/kube-scheduler.yaml"
  39. [upgrade/staticpods] Waiting for the kubelet to restart the component
  40. [apiclient] Found 1 Pods for label selector component=kube-scheduler
  41. [upgrade/staticpods] Component "kube-scheduler" upgraded successfully!
  42. [uploadconfig] Storing the configuration used in ConfigMap "kubeadm-config" in the "kube-system" Namespace
  43. [bootstraptoken] Configured RBAC rules to allow Node Bootstrap tokens to post CSRs in order for nodes to get long term certificate credentials
  44. [bootstraptoken] Configured RBAC rules to allow the csrapprover controller automatically approve CSRs from a Node Bootstrap Token
  45. [addons] Applied essential addon: kube-dns
  46. [addons] Applied essential addon: kube-proxy
  47.  
  48. [upgrade/successful] SUCCESS! Your cluster was upgraded to "v1.8.0". Enjoy!
  49.  
  50. [upgrade/kubelet] Now that your control plane is upgraded, please proceed with upgrading your kubelets in turn.

kubeadm upgrade apply 将执行下列步骤：

  • 检查集群是否处于可升级状态，包括：
    • API Server 是否可达，
    • 所有节点是否均处于 Ready 状态，并且
    • 控制平面处于健康状态
  • 强制启用版本偏移策略（version skew policy）。
  • 保证控制平面镜像可用或可以拉取到机器上。
  • 升级控制平面组件，当任何一个组件启动失败时对升级操作进行回退。
  • 应用新的 kube-dns 和 kube-proxy 清单文件并强制启用所有创建的必要 RBAC 规则。
  • 手动升级软件定义网络（Software Defined Network，SDN）。当前您的容器网络接口提供商（Container Network Interface，CNI）可能有自己的升级指导。请查阅 插件 页面，找到您的 CNI 提供商并查看是否有必要的额外升级步骤。
  • 为自动证书轮换添加必要的 RBAC 权限。将来 kubeadm 将自动执行这个步骤。
  1. $ kubectl create clusterrolebinding kubeadm:node-autoapprove-certificate-rotation --clusterrole=system:certificates.k8s.io:certificatesigningrequests:selfnodeclient --group=system:nodes

升级 master 和 node 软件包

对于集群中的每个节点（以下称为 $HOST），请运行下列命令升级其 kubelet。

  • 准备节点以进行维护，将其标记为不可调度并移除工作负载：
  1. $ kubectl drain $HOST --ignore-daemonsets

在 master 节点执行这个命令时，预计会出现这个错误，并且可以安全地将其忽略（因为 master 节点上有 static pod 运行）：

  1. node "master" already cordoned
  2. error: pods not managed by ReplicationController, ReplicaSet, Job, DaemonSet or StatefulSet (use --force to override): etcd-kubeadm, kube-apiserver-kubeadm, kube-controller-manager-kubeadm, kube-scheduler-kubeadm
  • 使用 Linux 发行版特定的包管理器升级 $HOST 节点上的 Kubernetes 软件包版本：
    如果节点运行基于 Debian 的发行版（如 Ubuntu），请执行：
  1. $ apt-get update
  2. $ apt-get upgrade

如果节点运行 CentOS 或类似发行版，请执行：

  1. $ yum update

现在，节点上应该运行的是新版本的 kubelet。请在 $HOST 上执行下列命令对此进行验证：

  1. $ systemctl status kubelet
  • 将节点标记为可调度（schedulable）以使其上线：
  1. $ kubectl uncordon $HOST
  • 在对所有集群节点的 kubelet 进行升级之后，请执行以下命令以确认所有节点又重新变为可用状态（从任何地方，例如集群外部）：
  1. $ kubectl get nodes

如果上述命令结果中所有节点的 STATUS 列都显示为 Ready，升级工作就已成功完成。

从损坏状态恢复

如果 kubeadm upgrade 因某些原因失败并且不能回退（可能因为执行过程中意外的关闭了节点实例），您可以再次运行 kubeadm upgrade，因为其具有幂等性，所以最终应该能够保证集群的实际状态和您所定义的理想状态一致。

您可以使用 kubeadm upgrade 命令和 x.x.x –> x.x.x 及 –force 参数，以从损坏状态恢复。

译者：xiaosuiba 原文链接

K8S中文社区微信公众号

原文: http://docs.kubernetes.org.cn/765.html