The following components will be restarted during the update from the Deckhouse 1.31
* Kubernetes Control Plane components of versions 1.19, 1.21, and 1.22
* `etcd`
* `kube-proxy`
Component version updates
* `cert-manager`: 1.7.1
* Kubernetes Control Plane: 1.21.11, 1.22.8
* `vector` (`log-shipper`): 0.20.0
Features
- For clusters with automatic Kubernetes version selection, Kubernetes v1.21 becomes the default version.
- The Kubernetes version update in such clusters will be done during the Deckhouse version update.
Updating the Kubernetes version will cause the cluster Control Plane components to restart.
- Run the following command to find out if the cluster has enabled automatic Kubernetes version selection:
shell
kubectl -n kube-system get secret d8-cluster-configuration -o json |
jq '.data."cluster-configuration.yaml"' -r | base64 -d | grep kubernetesVersion
If the result is `kubernetesVersion: Automatic` — the cluster has enabled automatic Kubernetes version selection.
- Run the following command in the cluster which has enabled automatic Kubernetes version selection to freeze the Kubernetes version to 1.19 before the update to Deckhouse v1.32:
shell
data=$(kubectl -n kube-system get secret d8-cluster-configuration -o json |
jq '.data."cluster-configuration.yaml"' -r | base64 -d | sed 's/kubernetesVersion: \"Automatic\"/kubernetesVersion: \"1.19\"/g' | base64 -w0)
kubectl -n kube-system patch secret d8-cluster-configuration -p "{\"data\": {\"cluster-configuration.yaml\": \"$data\"}}"
- The new module — [сeph-csi](https://deckhouse.io/en/documentation/v1.32.0/modules/099-ceph-csi/). Manages the creation of Ceph volumes (RBD and CephFS) and attaches them to workloads.
- The new module — [snapshot-controller](https://deckhouse.io/en/documentation/v1.32.0/modules/045-snapshot-controller/). Enables snapshot support for compatible CSI drivers and cloud providers in a Kubernetes cluster.
- The [linstor](https://deckhouse.io/en/documentation/v1.32.3/modules/031-linstor/) module improvements:
- Added Grafana dashboard to monitor LINSTOR cluster and DRBD resources.
- High availability mode support. LINSTOR components automatically switch to high availability mode in multi-master clusters.
- LINSTOR component monitoring improvements. Added metrics and alerts, including a description of steps to resolve a problem.
- LVM pools can automatically be imported to the LINSTOR cluster and StorageClasses are generated for them.
- The [openvpn](https://deckhouse.io/en/documentation/v1.32.3/modules/500-openvpn/) module got a new web interface and a significant improvement — it now doesn’t need persistent storage since it stores all necessary data in Secrets (automatic migration of existing configurations is provided). It is also possible to publish an openvpn Pod port using the `HostPort` inlet.
- Ingress Nginx controller version 1.1. Upgrading to this version will be necessary to upgrade to Kubernetes version 1.22 and higher.
- The [log-shipper](https://deckhouse.io/en/documentation/v1.32.3/modules/460-log-shipper/) module improvements:
- Namespace filtering in the `ClusterLoggingConfig` resource. You can specify both a list of namespaces to collect logs from and a list of namespaces to exclude.
- VPA mode is set to Initial. The log-shipper resources consumption depends on the number of Pods from which it collects logs. This number may differ from one node to another. But, VPA counts the average resource consumption across all nodes in the cluster, and it can lead to log-shipper Pods restart.
- NodeSelector and Tolerations options for the `log-shipper` agent Pods.
- Updated Grafana dashboard.
- Added [troubleshooting guide](https://deckhouse.io/en/documentation/v1.32.0/modules/460-log-shipper/advanced_usage.html).
See [the CHANGELOG v1.32](https://github.com/deckhouse/deckhouse/blob/main/CHANGELOG/CHANGELOG-v1.32.md) for more details.