Releases: kubermatic/kubermatic
Releases · kubermatic/kubermatic
v2.23.11
Bugfixes
- ACTION REQUIRED: For velero helm chart upgrade related change. If you use
velero.restic.deploy: true
, you will see new daemonsetnode-agent
running invelero
namespace. You might need to remove existing daemonset namedrestic
manually (#12998) - Fix panic, if no KubeVirt DNS config was set in the datacenter (#13029)
Updates
v2.22.13
v2.25.0 (Beta 0)
v2.24.2
Action Required
- ACTION REQUIRED: User Cluster MLA
cortex
chart has been upgraded to resolve issues for cortex-compactor and improve stability of the User Cluster MLA feature. Few actions are required to be taken to use new upgraded charts (#12935)- Refer to Upstream helm chart values to see the latest default values
- Some of the values from earlier
values.yaml
are now incompatible with latest version. They are removed in thevalues.yaml
in the current chart. But if you had copied the original values.yaml to customize it further, you may see thatkubermatic-installer
will detect such incompatible options and churn out errors and explain that action that needs to be taken. - The memcached-* charts are now subcharts of cortex chart so if you provided configuration for
memcached-*
blocks in yourvalues.yaml
for user-mla, you must move them undercortex:
block
Updates
- Add support for Kubernetes v1.26.13, v1.27.10, v1.28.6 and set default version to v1.27.10 (#12982)
Bugfixes
- If the seed cluster is using Cilium as CNI, create CiliumClusterwideNetworkPolicy for api-server connectivity (#12966)
- Stop constantly re-deploying operating-system-manager when registry mirrors are configured (#12972)
- The Kubermatic installer will now detect DNS settings based on the Ingress instead of the nginx-ingress LoadBalancer, allowing for other ingress solutions to be properly detected (#12934)
Removals
- Remove 1.25 from list of supported versions on AKS (EOL on January 14th) (#12962)
v2.23.10
Bugfixes
- Stop constantly re-deploying operating-system-manager when registry mirrors are configured (#12972)
Updates
- Update EKS/AKS version matrix to only include Kubernetes versions supported by those managed offerings. For AKS 1.26-1.28 are supported, for EKS 1.24 to 1.28. The default for newly created external clusters is now 1.28 (#12964)
- Add support for Kubernetes v1.24.17, v1.25.16, v1.26.13, v1.27.10 and set default version to v1.26.13 (#12983)
v2.22.12
Bugfixes
- Stop constantly re-deploying operating-system-manager when registry mirrors are configured (#12972)
Updates
- Update EKS/AKS version matrix to only include Kubernetes versions supported by those managed offerings. For AKS 1.26-1.28 are supported, for EKS 1.24 to 1.28. The default for newly created external clusters is now 1.28 (#12965)
- Add support for Kubernetes v1.24.17, v1.25.16 and v1.26.13 and set default version to v1.25.16 (#12984)
v2.24.1
Bugfixes
- Applied a fix to VPA caused by upstream release issue which caused insufficient RBAC permission for VPA recommender pod (#12872)
- Fix cert-manager values block. cert-manager deployment will get updated as part of upgrade (#12854)
- Fix
mirror-images
command in installer not being able to extract the addons (#12868) - Fix cases where, when using dedicated infra- and ccm-credentials, infra-credentials were always overwritten by ccm-credentials (#12421)
- No longer fail constructing vSphere endpoint when a
/
suffix is present in the datacenter configuration (#12861)
New Features
- Openstack: allow configuring Cinder CSI topology support either on
Cluster
orSeed
resource fieldcinderTopologyEnabled
(#12878)
Updates
- Update machine-controller to v1.58.1 (#12902)
- Update Anexia CCM (cloud-controller-manager) to version 1.5.5 (#12911)
- Fixes leaking LoadBalancer reconciliation metric
- Updates various dependencies
Miscellaneous
v2.23.9
Bugfixes
- Applied a fix to VPA caused by upstream release issue which caused insufficient RBAC permission for VPA recommender pod (#12872)
- Fix cert-manager values block. cert-manager deployment will get updated as part of upgrade (#12854)
- Fix cases where, when using dedicated infra- and ccm-credentials, infra-credentials were always overwritten by ccm-credentials (#12421)
- No longer fail constructing vSphere endpoint when a
/
suffix is present in the datacenter configuration (#12861)
Updates
- Update machine-controller to v1.57.4 (#12903)
- Update Anexia CCM (cloud-controller-manager) to version 1.5.5 (#12910)
- Fixes leaking LoadBalancer reconciliation metric
- Updates various dependencies