2.6.0 - Update Kubernetes to 1.25.9

This release mostly updates Kubernetes and components to newer versions.

• Additionally the OS image template upload has been changed from uploading an OVA to an OVF URL. The image URL provided by default points to an OVF with contains an ExtraConfig parameter disk.EnableUUID=true, to allow use of the vCloud-CSI by the user if requested.

• ⚠️ Breaking Change: If Helm has problems upgrading an existing Longhorn from a previous version, make sure to delete the Helm configuration secrets in the longhorn-system namespace first:

$ kubectl -n longhorn-system get secret
NAME                                    TYPE                 DATA   AGE
secret/longhorn-webhook-ca              kubernetes.io/tls    2      237d
secret/longhorn-webhook-tls             kubernetes.io/tls    2      237d
secret/sh.helm.release.v1.longhorn.v1   helm.sh/release.v1   1      237d
secret/sh.helm.release.v1.longhorn.v2   helm.sh/release.v1   1      170d

$ kubectl -n longhorn-system delete secret/sh.helm.release.v1.longhorn.v1 secret/sh.helm.release.v1.longhorn.v2
secret "sh.helm.release.v1.longhorn.v1" deleted
secret "sh.helm.release.v1.longhorn.v2" deleted

---

Component updates:

Component	New version
k3s	v1.25.9+k3s1
helm-longhorn	1.4.1
helm-kubernetes-dashboard	5.11.0

---

⚠️ Warning: This version upgrades Kubernetes to v1.25.x, in which PodSecurityPolicy support has been removed. If you have any deployments that make use of PSP's (or you want to avoid any potential hassle with the included Longhorn upgrade), then you might want to stay on the previous release and consider not updating.

2.5.0 - Update Kubernetes to 1.24.10

• Breaking Change: Be aware that this release will switch over to a much newer prometheus chart. Please scale down the prometheus-server deployment to 0 before you upgrade your K8s cluster, and also delete the existing prometheus-pushgateway:
  ⚠️ kubectl -n prometheus scale deploy prometheus-server --replicas=0
  ⚠️ kubectl -n prometheus delete deploy prometheus-pushgateway
  
  If you do not do this then your Helm chart deployment might fail and will need to be fixed manually!
  See https://artifacthub.io/packages/helm/prometheus-community/prometheus#to-19-0 for further information.

Component updates:

Component	New version
k3s	v1.24.10+k3s1
helm-cert-manager	1.11.0
helm-ingress-controller	4.4.2
helm-kured	4.4.1
helm-loki	4.6.1
helm-prometheus	19.6.0
helm-promtail	6.8.3

2.4.2 - Update Kubernetes to 1.24.8

This release updates Kubernetes and components to newer versions.

Component updates:

Component	New version
k3s	v1.24.8+k3s1
cilium	v1.12.4
helm-cert-manager	1.10.1
helm-promtail	6.6.2

2.4.1 - Fix version annotations

This release fixes a bug in version/release annotations on nodes. If you deployed a version newer than v2.3.0 it would fail on a fresh, new deployment.

2.4.0 - Update Kubernetes to 1.24.7

• Breaking Change: Be aware that this release will switch over to a much newer Loki Helm chart. Unfortunately due to a change in the architecture the Loki deployment might have its PersistentVolume being replaced, this means you would lose all previously stored logfiles. In order to minimize downtime and possible issues, please delete the Loki StatefulSet before you upgrade your K8s cluster:
  ⚠️ kubectl -n loki delete statefulsets.apps loki
  If you do not do this then your Helm chart deployment might fail and will need to be fixed manually!
  See https://grafana.com/docs/loki/latest/installation/helm/upgrade-from-2.x/ for further information.

Component updates:

Component	New version
k3s	v1.24.7+k3s1
cilium	v1.12.3
cilium-cli	v0.12.6
helm-longhorn	1.3.2
helm-kured	4.1.0
helm-ingress-nginx	4.4.0
helm-cert-manager	1.10.0
helm-prometheus	15.18.0
helm-loki	3.3.4
helm-promtail	6.6.1
helm-grafana	6.43.5

Note: If the cilium installation gets stuck during the upgrade, check the status of all pods on your cluster with kubectl get pods -A. Particularly prometheus, grafana and loki might get stuck in status Terminating, due to Kubernetes failing to detach the volumes from these pods. This can happen because as part of a cilium upgrade all pods on the cluster get restarted, also the ones from Longhorn which is supposed to manage their PVC/PVs.

2.3.1 - Add release version to K3s deployment

Adds the release version (git tag) to the Kubernetes deployment. It will add version and commit hash as annotations to all control plane and worker nodes. This should make it easier for users to track and see which version is currently deployed.

2.3.0 - Unattended-Upgrades and Node reboots

This release adds two new configuration options to variables.tf:

Variable	Default	Description
k8s_automatically_upgrade_os	true	Enables unattended-upgrades on Ubuntu OS VMs
k8s_enable_automatic_node_reboot	true	Enables automatic node reboots via kured for OS upgrades

The deployed kured daemonset is configured to allow reboots only on Monday, Tuesday, Wednesday and Thursday, between 02:00 and 05:00 UTC.

If you do not want unattended-upgrades of packages on your VMs, then you should set k8s_automatically_upgrade_os to false.
If you do not want your Kubernetes nodes to reboot and/or your pods cannot handle being rescheduled properly, then you should set k8s_enable_automatic_node_reboot to false.

2.2.1 - Configurable ACME directory server

Adds a new configuration variable: k8s_cert_manager_lets_encrypt_server.
With the default value being set to https://acme-v02.api.letsencrypt.org/directory, there are no changes expected to be done on user side.

2.2.0 - Update Kubernetes to 1.24.6

This release updates Kubernetes and components to newer versions, and also adds a Sonobuoy quick run to the GitHub workflow.

Component updates:

Component	New version
k3s	v1.24.6+k3s1
cilium	v1.12.2
cilium-cli	v0.12.4
helm-ingress-nginx	4.2.5
helm-prometheus	15.12.2
helm-loki	2.16.0
helm-promtail	6.4.0
helm-grafana	6.39.0

2.1.2 - LoadBalancer dependency

Fixes a small dependency bug between VMs / Kubernetes and the LoadBalancer configuration.