0.9.4

[gardener-extensions]

Improvements

• [USER] The CSI controllers of Alicloud and Packet are now correctly hibernated and woken up during deletion of hibernated and non-hibernated shoots. (55a8d98)## Docker Images
  gardener-extension-hyper: eu.gcr.io/gardener-project/gardener/gardener-extension-hyper:0.9.4

0.9.3

[gardener-extensions]

Improvements

• [OPERATOR] The certificate secrets stored by webhooks are no longer updated if they have not been found. This prevents that multiple pods are mutually overwritten each others freshly generated certificates and start serving an different ones. (f41e114)
• [OPERATOR] Webhooks are no storing their certificates inside the Kubernetes cluster they are acting on. It's not needed to regenerate them at every restart. (715b35a)
• [OPERATOR] The generic control plane actuator does now correctly wake up provider-specific control plane components when a hibernated shoot is deleted. (882ac8a)
• [OPERATOR] The OpenStack control plane actuator is no longer wrapped because the deletion problem is fixed more generally with gardener/gardener#1317. (5ba64f4)

Docker Images

gardener-extension-hyper: eu.gcr.io/gardener-project/gardener/gardener-extension-hyper:0.9.3

0.9.2

[gardener-extensions]

Improvements

• [USER] An issue that caused AWS infrastructure controller to not properly clean up remaining Kubernetes load balancers and security groups has been fixed (only occurred when the AWS account had over 400 load balancers). (d148aa4)
• [OPERATOR] Upgrade AliCloud CSI diskplugin to latest version v1.13.2-3 (b17c3f8)

Docker Images

gardener-extension-hyper: eu.gcr.io/gardener-project/gardener/gardener-extension-hyper:0.9.2

0.9.1

[gardener-extensions]

Improvements

• [OPERATOR] A bug in the OS controllers that resulted in wrongly generated cloud-init scripts has been fixed. The Ubuntu and the SUSE-JeOS controllers are now generating the correct config again. (02c2ad8)

Docker Images

gardener-extension-hyper: eu.gcr.io/gardener-project/gardener/gardener-extension-hyper:0.9.1

0.9.0

[gardener-extensions]

Most notable changes

• [USER] provider-aws sets the hostname of the VMs to be their FQDN. (#236, @vpnachev)
• [USER] MTU values on AWS nodes are now changed to 1460 instead of 9000. (#228, @zanetworker)
• [OPERATOR] Added LoadBalancer Classes support for OpenStack Cloud Provider (#222, @afritzler)
• [DEVELOPER] The controller-runtime library has been upgraded to 0.2.0-beta.2. (#218, @rfranzke)

Improvements

• [USER] Updates worst-case machine-drain-timeout to 12hrs (#229, @prashanth26)
• [USER] Upgrade AliCloud CSI diskplugin to latest version v1.13.2-1 (#221, @jia-jerry)
• [OPERATOR] The default value of the ignoreOperationAnnotation flag has been changed from true to false. (#234, @timuthy)
• [OPERATOR] The os-ubuntu controller was configured to ignore the operation annotation which didn't play well with the latest Gardener release. This problem is now fixed. (#225, @vpnachev)
• [OPERATOR] The amount of required reconciliations for Backupbucket and BackupEntry controllers has been improved. (#224, @timuthy)
• [DEVELOPER] The Certificate-Service now uses ManagedResources instead of applying ClusterRoles and ClusterRolebindings directly to shoot clusters. (#234, @timuthy)
• [DEVELOPER] Updates MCM ClusterRole for shoot to allow access to PV and PVC resources (#229, @prashanth26)

[machine-controller-manager]

Most notable changes

• [USER] MCM now needs permissions to GET/LIST PV and PVCs on target cluster while draining machines (gardener/machine-controller-manager#275, @ggaurav10)
• [OPERATOR] Changes the drain logic to honour the individual pods' terminationGracePeriodSeconds instead of a fixed(though configurable) timeout. For the pods without volumes, the timeout would be a bit longer than the maximum of terminationGracePeriodSeconds of the pods without volume. For the pods with volumes, the timeout is its own terminationGracePeriodSeconds and a fixed (but configurable) volume detach timeout for each pod. (gardener/machine-controller-manager#288, @amshuman-kr)
  • The command-line flags machine-max-evict-retries and machine-pv-detach-timeout can be used to customize the default behaviour of the number of retries for eviction and wait period for the detaching of volumes after pods are evicted/deleted.
  • The default value for the existing command-line flag machine-drain-timeout has been increased to 12h which should be enough for most work-loads.

Improvements

• [OPERATOR] Bugfix: Allows force deletion of pods who have no controllers backing them (gardener/machine-controller-manager#301, @prashanth26)
• [OPERATOR] Increased OpenStack server status wait for a timeout during server creation from 5 to 10 mins. (gardener/machine-controller-manager#286, @kayrus)
• [OPERATOR] Improved the server status handling - It no longer waits for the timeout if the server status is different from BUILD during server creation. (gardener/machine-controller-manager#286, @kayrus)
• [OPERATOR] Force deletion of machine succeeds even on drain failures (gardener/machine-controller-manager#283, @prashanth26)
• [OPERATOR] dep is replaced by go mod. (gardener/machine-controller-manager#280, @ialidzhikov)
• [OPERATOR] When draining a node, pods with PVCs are evicted serially. Next pod eviction waits for PV of the previously evicted pod to detach from the node. Operators can expect faster machine drains for their machines. (gardener/machine-controller-manager#275, @ggaurav10)

Docker Images

gardener-extension-hyper: eu.gcr.io/gardener-project/gardener/gardener-extension-hyper:0.9.0

0.8.0

[gardener-extensions]

Action Required

• [DEVELOPER] The WorkerDelegate interface does now have new GetMachineImages() interface method. Provider extension that are using the generic worker actuator have to implement this. It should return a provider-specific WorkerStatus type that will get written to the Worker's .status.providerStatus field. The worker delegate can use this information when computing the machine classes to try to find the provider-specific machine image data for the desired version. This is useful for cases where a version was removed (or updated) from the componentconfig of the provider extension, but old Worker resource still use this version. Without this information the worker controller would not be able to reconcile anymore. (#217, @rfranzke)

Most notable changes

• [USER] The provider extension controllers do now support Kubernetes v1.15.0. (129567e)
• [OPERATOR] The Helm charts of every extension controller does now contain a VPA configuration for automated vertical scaling. (#197, @georgekuruvillak)
• [OPERATOR] For 1.15 Azure clusters the cloud-config used by kubelet does no longer contain credentials. (#192, @zanetworker)
• [OPERATOR] All extension controllers do now only reconcile their extension resources if the gardener.cloud/operation=reconcile annotation is set in the respective resource. Beginning with Gardener version 0.27.0 the gardener-controller-manager will annotate these extension resources when reconciling a shoot. This is to give Gardener control when the extension controllers should act (e.g., only in the shoot's maintenance time window). All extension controllers must implement this contract (see also this document). (#178, @timuthy)
  • ⚠️ Due to this change this version of the extensions is not compatible with Gardener versions older than 0.27.0.
• [OPERATOR] Added a new operating system config controller that supports Ubuntu OS. (#103, @vlvasilev)

Improvements

• [USER] Deleting a shoot on GCP will now only delete the routes that shoot when multiple shoots share the same VPC network. (#170, @plkokanov)
• [USER] Deleting a shoot on GCP will now only delete the firewall rules for that shoot when multiple shoots share the same VPC network. (#169, @plkokanov)
• [OPERATOR] The generic worker actuator does now re-apply the machine-controller-manager deployment during worker deletion to ensure a proper worker node deletion. (#215, @timuthy)
• [OPERATOR] The generic worker actuator is now scaling down and up the cluster-autoscaler instead of deleting the deployment (and relying on Gardener to re-deploy it). (#211, @rfranzke)
• [OPERATOR] The provider extension controllers for Alicloud does now support Kubernetes v1.15.0. (#198, @EmoinLanyu)
• [OPERATOR] Added a 4 minutes delay before removing the OpenStack cloud controller manager when deleting a control plane. This is to give the CCM enough time to cleanup load balancers and other resources in the infrastructure. With service finalizers graduated to beta (Kubernetes 1.16) this can be removed again. (#191, @afritzler)
• [OPERATOR] The Helm charts of all provider extensions have been improved to not produce checksum annotations that are too long. (#174, @jia-jerry)
• [OPERATOR] The Alicloud extension Helm chart does now support using the flexvolume provider. (#172, @jia-jerry)
• [OPERATOR] Garden-extension now supports the backup-extension resource for AWS provider. (#123, @swapnilgm)
• [OPERATOR] The control plane actuators do now create ManagedResources for storage classes. (9ed92be)
• [OPERATOR] The control plane actuators do now create ManagedResources for storage classes. (38e7ab4)
• [OPERATOR] The RBAC roles for the machine-controller-manager in the shoot are now managed via ManagedResources created by the worker actuators. (5102585)

Docker Images

gardener-extension-hyper: eu.gcr.io/gardener-project/gardener/gardener-extension-hyper:0.8.0

0.7.4

no release notes available

Docker Images

gardener-extension-hyper: eu.gcr.io/gardener-project/gardener/gardener-extension-hyper:0.7.4

0.7.3

[gardener-extensions]

Improvements

• [OPERATOR] The Helm charts of all provider extensions have been improved to not produce checksum annotations that are too long. (#177, @jia-jerry)
• [OPERATOR] The Alicloud extension Helm chart does now support using the flexvolume provider. (#176, @jia-jerry)

Docker Images

gardener-extension-hyper: eu.gcr.io/gardener-project/gardener/gardener-extension-hyper:0.7.3

0.7.2

[gardener-extensions]

Improvements

• [USER] Deleting a shoot on GCP will now only delete the routes that shoot when multiple shoots share the same VPC network. (25735f9)
• [USER] Deleting a shoot on GCP will now only delete the firewall rules for that shoot when multiple shoots share the same VPC network. (f40c51b)

Docker Images

gardener-extension-hyper: eu.gcr.io/gardener-project/gardener/gardener-extension-hyper:0.7.2

0.7.1

[gardener-extensions]

Most notable changes

• [USER] The provider extension controllers for AWS, Azure, GCP, and OpenStack do now support Kubernetes v1.15.0. (10d3f10)

Docker Images

gardener-extension-hyper: eu.gcr.io/gardener-project/gardener/gardener-extension-hyper:0.7.1