|
| 1 | +--- |
| 2 | +reviewers: |
| 3 | +- sig-cluster-lifecycle |
| 4 | +title: Reconfiguring a kubeadm cluster |
| 5 | +content_type: task |
| 6 | +weight: 10 |
| 7 | +--- |
| 8 | + |
| 9 | +<!-- overview --> |
| 10 | + |
| 11 | +kubeadm does not support automated ways of reconfiguring components that |
| 12 | +were deployed on managed nodes. One way of automating this would be |
| 13 | +by using a custom [operator](/docs/concepts/extend-kubernetes/operator/). |
| 14 | + |
| 15 | +To modify the components configuration you must manually edit associated cluster |
| 16 | +objects and files on disk. |
| 17 | + |
| 18 | +This guide shows the correct sequence of steps that need to be performed |
| 19 | +to achieve kubeadm cluster reconfiguration. |
| 20 | + |
| 21 | +## {{% heading "prerequisites" %}} |
| 22 | + |
| 23 | +- You need a cluster that was deployed using kubeadm |
| 24 | +- Have administrator credentials (`/etc/kubernetes/admin.conf`) and network connectivity |
| 25 | +to a running kube-apiserver in the cluster from a host that has kubectl installed |
| 26 | +- Have a text editor installed on all hosts |
| 27 | + |
| 28 | +<!-- steps --> |
| 29 | + |
| 30 | +## Reconfiguring the cluster |
| 31 | + |
| 32 | +kubeadm writes a set of cluster wide component configuration options in |
| 33 | +ConfigMaps and other objects. These objects must be manually edited. The command `kubectl edit` |
| 34 | +can be used for that. |
| 35 | + |
| 36 | +The `kubectl edit` command will open a text editor where you can edit and save the object directly. |
| 37 | + |
| 38 | +You can use the environment variables `KUBECONFIG` and `KUBE_EDITOR` to specify the location of |
| 39 | +the kubectl consumed kubeconfig file and preferred text editor. |
| 40 | + |
| 41 | +For example: |
| 42 | +``` |
| 43 | +KUBECONFIG=/etc/kubernetes/admin.conf KUBE_EDITOR=nano kubectl edit <parameters> |
| 44 | +``` |
| 45 | + |
| 46 | +{{< note >}} |
| 47 | +Upon saving any changes to these cluster objects, components running on nodes may not be |
| 48 | +automatically updated. The steps below instruct you on how to perform that manually. |
| 49 | +{{< /note >}} |
| 50 | + |
| 51 | +{{< warning >}} |
| 52 | +Component configuration in ConfigMaps is stored as unstructured data (YAML string). |
| 53 | +This means that validation will not be performed upon updating the contents of a ConfigMap. |
| 54 | +You have to be careful to follow the documented API format for a particular |
| 55 | +component configuration and avoid introducing typos and YAML indentation mistakes. |
| 56 | +{{< /warning >}} |
| 57 | + |
| 58 | +### Applying cluster configuration changes |
| 59 | + |
| 60 | +#### Updating the `ClusterConfiguration` |
| 61 | + |
| 62 | +During cluster creation and upgrade, kubeadm writes its |
| 63 | +[`ClusterConfiguration`](/docs/reference/config-api/kubeadm-config.v1beta3/) |
| 64 | +in a ConfigMap called `kubeadm-config` in the `kube-system` namespace. |
| 65 | + |
| 66 | +To change a particular option in the `ClusterConfiguration` you can edit the ConfigMap with this command: |
| 67 | + |
| 68 | +```shell |
| 69 | +kubectl edit cm -n kube-system kubeadm-config |
| 70 | +``` |
| 71 | + |
| 72 | +The configuration is located under the `data.ClusterConfiguration` key. |
| 73 | + |
| 74 | +{{< note >}} |
| 75 | +The `ClusterConfiguration` includes a variety of options that affect the configuration of individual |
| 76 | +components such as kube-apiserver, kube-scheduler, kube-controller-manager, CoreDNS, etcd and kube-proxy. |
| 77 | +Changes to the configuration must be reflected on node components manually. |
| 78 | +{{< /note >}} |
| 79 | + |
| 80 | +#### Reflecting `ClusterConfiguration` changes on control plane nodes |
| 81 | + |
| 82 | +kubeadm manages the control plane components as static Pod manifests located in |
| 83 | +the directory `/etc/kubernetes/manifests`. |
| 84 | +Any changes to the `ClusterConfiguration` under the `apiServer`, `controllerManager`, `scheduler` or `etcd` |
| 85 | +keys must be reflected in the associated files in the manifests directory on a control plane node. |
| 86 | + |
| 87 | +Such changes may include: |
| 88 | +- `extraArgs` - requires updating the list of flags passed to a component container |
| 89 | +- `extraMounts` - requires updated the volume mounts for a component container |
| 90 | +- `*SANs` - requires writing new certificates with updated Subject Alternative Names. |
| 91 | + |
| 92 | +Before proceeding with these changes, make sure you have backed up the directory `/etc/kubernetes/`. |
| 93 | + |
| 94 | +To write new certificates you can use: |
| 95 | +```shell |
| 96 | +kubeadm init phase certs <component-name> --config <config-file> |
| 97 | +``` |
| 98 | + |
| 99 | +To write new manifest files in `/etc/kubernetes/manifests` you can use: |
| 100 | + |
| 101 | +```shell |
| 102 | +kubeadm init phase control-plane <component-name> --config <config-file> |
| 103 | +``` |
| 104 | + |
| 105 | +The `<config-file>` contents must match the updated `ClusterConfiguration`. |
| 106 | +The `<component-name>` value must be the name of the component. |
| 107 | + |
| 108 | +{{< note >}} |
| 109 | +Updating a file in `/etc/kubernetes/manifests` will tell the kubelet to restart the static Pod for the corresponding component. |
| 110 | +Try doing these changes one node at a time to leave the cluster without downtime. |
| 111 | +{{< /note >}} |
| 112 | + |
| 113 | +### Applying kubelet configuration changes |
| 114 | + |
| 115 | +#### Updating the `KubeletConfiguration` |
| 116 | + |
| 117 | +During cluster creation and upgrade, kubeadm writes its |
| 118 | +[`KubeletConfiguration`](/docs/reference/config-api/kubelet-config.v1beta1/) |
| 119 | +in a ConfigMap called `kubelet-config` in the `kube-system` namespace. |
| 120 | + |
| 121 | +You can edit the ConfigMap with this command: |
| 122 | + |
| 123 | +```shell |
| 124 | +kubectl edit cm -n kube-system kubelet-config |
| 125 | +``` |
| 126 | + |
| 127 | +The configuration is located under the `data.kubelet` key. |
| 128 | + |
| 129 | +#### Reflecting the kubelet changes |
| 130 | + |
| 131 | +To reflect the change on kubeadm nodes you must do the following: |
| 132 | +- Log in to a kubeadm node |
| 133 | +- Run `kubeadm upgrade node phase kubelet-config` to download the latest `kubelet-config` |
| 134 | +ConfigMap contents into the local file `/var/lib/kubelet/config.conf` |
| 135 | +- Edit the file `/var/lib/kubelet/kubeadm-flags.env` to apply additional configuration with |
| 136 | +flags |
| 137 | +- Restart the kubelet service with `systemctl restart kubelet` |
| 138 | + |
| 139 | +{{< note >}} |
| 140 | +Do these changes one node at a time to allow workloads to be rescheduled properly. |
| 141 | +{{< /note >}} |
| 142 | + |
| 143 | +{{< note >}} |
| 144 | +During `kubeadm upgrade`, kubeadm downloads the `KubeletConfiguration` from the |
| 145 | +`kubelet-config` ConfigMap and overwrite the contents of `/var/lib/kubelet/config.conf`. |
| 146 | +This means that node local configuration must be applied either by flags in |
| 147 | +`/var/lib/kubelet/kubeadm-flags.env` or by manually updating the contents of |
| 148 | +`/var/lib/kubelet/config.conf` after `kubeadm upgrade`, and then restarting the kubelet. |
| 149 | +{{< /note >}} |
| 150 | + |
| 151 | +### Applying kube-proxy configuration changes |
| 152 | + |
| 153 | +#### Updating the `KubeProxyConfiguration` |
| 154 | + |
| 155 | +During cluster creation and upgrade, kubeadm writes its |
| 156 | +[`KubeProxyConfiguration`](/docs/reference/config-api/kube-proxy-config.v1alpha1/) |
| 157 | +in a ConfigMap in the `kube-system` namespace called `kube-proxy`. |
| 158 | + |
| 159 | +This ConfigMap is used by the `kube-proxy` DaemonSet in the `kube-system` namespace. |
| 160 | + |
| 161 | +To change a particular option in the `KubeProxyConfiguration`, you can edit the ConfigMap with this command: |
| 162 | + |
| 163 | +```shell |
| 164 | +kubectl edit cm -n kube-system kube-proxy |
| 165 | +``` |
| 166 | + |
| 167 | +The configuration is located under the `data.config.conf` key. |
| 168 | + |
| 169 | +#### Reflecting the kube-proxy changes |
| 170 | + |
| 171 | +Once the `kube-proxy` ConfigMap is updated, you can restart all kube-proxy Pods: |
| 172 | + |
| 173 | +Obtain the Pod names: |
| 174 | + |
| 175 | +```shell |
| 176 | +kubectl get po -n kube-system | grep kube-proxy |
| 177 | +``` |
| 178 | + |
| 179 | +Delete a Pod with: |
| 180 | + |
| 181 | +```shell |
| 182 | +kubectl delete po -n kube-system <pod-name> |
| 183 | +``` |
| 184 | + |
| 185 | +New Pods that use the updated ConfigMap will be created. |
| 186 | + |
| 187 | +{{< note >}} |
| 188 | +Because kubeadm deploys kube-proxy as a DaemonSet, node specific configuration is unsupported. |
| 189 | +{{< /note >}} |
| 190 | + |
| 191 | +### Applying CoreDNS configuration changes |
| 192 | + |
| 193 | +#### Updating the CoreDNS Deployment and Service |
| 194 | + |
| 195 | +kubeadm deploys CoreDNS as a Deployment called `coredns` and with a Service `kube-dns`, |
| 196 | +both in the `kube-system` namespace. |
| 197 | + |
| 198 | +To update any of the CoreDNS settings, you can edit the Deployment and |
| 199 | +Service objects: |
| 200 | + |
| 201 | +```shell |
| 202 | +kubectl edit deployment -n kube-system coredns |
| 203 | +kubectl edit service -n kube-system kube-dns |
| 204 | +``` |
| 205 | + |
| 206 | +#### Reflecting the CoreDNS changes |
| 207 | + |
| 208 | +Once the CoreDNS changes are applied you can delete the CoreDNS Pods: |
| 209 | + |
| 210 | +Obtain the Pod names: |
| 211 | + |
| 212 | +```shell |
| 213 | +kubectl get po -n kube-system | grep coredns |
| 214 | +``` |
| 215 | + |
| 216 | +Delete a Pod with: |
| 217 | + |
| 218 | +```shell |
| 219 | +kubectl delete po -n kube-system <pod-name> |
| 220 | +``` |
| 221 | + |
| 222 | +New Pods with the updated CoreDNS configuration will be created. |
| 223 | + |
| 224 | +{{< note >}} |
| 225 | +kubeadm does not allow CoreDNS configuration during cluster creation and upgrade. |
| 226 | +This means that if you execute `kubeadm upgrade apply`, your changes to the CoreDNS |
| 227 | +objects will be lost and must be reapplied. |
| 228 | +{{< /note >}} |
| 229 | + |
| 230 | +## Persisting the reconfiguration |
| 231 | + |
| 232 | +During the execution of `kubeadm upgrade` on a managed node, kubeadm might overwrite configuration |
| 233 | +that was applied after the cluster was created (reconfiguration). |
| 234 | + |
| 235 | +### Persisting Node object reconfiguration |
| 236 | + |
| 237 | +kubeadm writes Labels, Taints, CRI socket and other information on the Node object for a particular |
| 238 | +Kubernetes node. To change any of the contents of this Node object you can use: |
| 239 | + |
| 240 | +```shell |
| 241 | +kubectl edit no <node-name> |
| 242 | +``` |
| 243 | + |
| 244 | +During `kubeadm upgrade` the contents of such a Node might get overwritten. |
| 245 | +If you would like to persist your modifications to the Node object after upgrade, |
| 246 | +you can prepare a [kubectl patch](/docs/tasks/manage-kubernetes-objects/update-api-object-kubectl-patch/) |
| 247 | +and apply it to the Node object: |
| 248 | + |
| 249 | +```shell |
| 250 | +kubectl patch no <node-name> --patch-file <patch-file> |
| 251 | +``` |
| 252 | + |
| 253 | +#### Persisting control plane component reconfiguration |
| 254 | + |
| 255 | +The main source of control plane configuration is the `ClusterConfiguration` |
| 256 | +object stored in the cluster. To extend the static Pod manifests configuration, |
| 257 | +[patches](/docs/setup/production-environment/tools/kubeadm/control-plane-flags/#patches) can be used. |
| 258 | + |
| 259 | +These patch files must remain as files on the control plane nodes to ensure that |
| 260 | +they can be used by the `kubeadm upgrade ... --patches <directory>`. |
| 261 | + |
| 262 | +If reconfiguration is done to the `ClusterConfiguration` and static Pod manifests on disk, |
| 263 | +the set of node specific patches must be updated accordingly. |
| 264 | + |
| 265 | +#### Persisting kubelet reconfiguration |
| 266 | + |
| 267 | +Any changes to the `KubeletConfiguration` stored in `/var/lib/kubelet/config.conf` will be overwritten on |
| 268 | +`kubeadm upgrade` by downloading the contents of the cluster wide `kubelet-config` ConfigMap. |
| 269 | +To persist kubelet node specific configuration either the file `/var/lib/kubelet/config.conf` |
| 270 | +has to be updated manually post-upgrade or the file `/var/lib/kubelet/kubeadm-flags.env` can include flags. |
| 271 | +The kubelet flags override the associated `KubeletConfiguration` options, but note that |
| 272 | +some of the flags are deprecated. |
| 273 | + |
| 274 | +A kubelet restart will be required after changing `/var/lib/kubelet/config.conf` or |
| 275 | +`/var/lib/kubelet/kubeadm-flags.env`. |
| 276 | + |
| 277 | +{{% heading "whatsnext" %}} |
| 278 | + |
| 279 | +- [Upgrading kubeadm clusters](/docs/tasks/administer-cluster/kubeadm/kubeadm-upgrade) |
| 280 | +- [Customizing components with the kubeadm API](/docs/setup/production-environment/tools/kubeadm/control-plane-flags) |
| 281 | +- [Certificate management with kubeadm](/docs/tasks/administer-cluster/kubeadm/kubeadm-certs) |
0 commit comments