0.6.0

CHANGELOG:

• Safety reconciliation is now split into different threads for each safety handlers (MachineObjectOvershooting, OrphanVMHandler)
• Lesser number of API calls to cloud provider to check for orphan VMs (configurable now)
• Informers for control core object is now filtered. Avoids unwanted events
• Add finaliser on secret objects referred by existing machineClasses
• Fixed bugs while supporting for AWS io1 volume type
• Better logging while creation/updation/deletion of machine objects

Docker image: eu.gcr.io/gardener-project/gardener/machine-controller-manager:0.6.0

0.5.0

CHANGELOG:

• Propagated machine failure status to machineSets and machineDeployments.
• Update object status for failed machine objects, this avoids re-creation of failed machines when machines joins after a long delay.
• Azure NIC and VM creation made into a transaction.
• Fixes bug while deploying machines backed by SSD on GCP.
• Freeze logic now allows scaling down of VMs, but not scale-up.
• Disabled MachineTimeout freeze.
• Fixed openstack multi image creation problem.
• Added first set of unit tests and enabled unit tests in CI tests.

Docker image: eu.gcr.io/gardener-project/gardener/machine-controller-manager:0.5.0

0.4.0

CHANGELOG:

• Machine-controller-manager now does a set of integration tests for AWS
• Updated notices and licences
• Replaced maintainers.md to codeowners
• Removed bug to enforce draining of nodes always
• Better error handling while creation of Azure machines

Docker image: eu.gcr.io/gardener-project/gardener/machine-controller-manager:0.4.0

0.3.0

CHANGELOG:

• MachineDeployment now prioritizes the Machines while scaling down. On top of prioritizing Unhealthy machines to be deleted first, it also checks the annotation machinepriority.machine.sapcloud.io , prefers in increasing order. commit reference

• Azure support has been extended to enable safety-controller, to keep track of NIC, VM, Disks separately. This is inspired by the attribute of Azure, where these resources need to be taken care of separately while creation/deletion.

• Removed the vendor dependency from k8s.io/kubernetes.

Docker image: eu.gcr.io/gardener-project/gardener/machine-controller-manager:0.3.0

0.2.0

CHANGELOG:

• Integrated safety controller with mcm.
  • Safety controller freezes mcm when a number of machine-objects/VMs crosses predefined number.
  • Safety controller also removes the orphan VMs if create by any chance.
  • It supports following infrastructures: aws, gcp, azure, openstack
• Change in the nomenclature of machine tags. Made certain tags compulsory.
• Respects the Application SLO via PDBs while rolling-update of MachinDeployment.
• Minor updates in the logic of machine-controller
  • Does not expect providerID on machine-object while deletion.
  • Does not reconcile if machine creation fails, leaves responsibility to MachineSet.
  • Allows deletion of Machine object if cloud-provider throws "Resource Not Found" error.

Docker image: eu.gcr.io/gardener-project/gardener/machine-controller-manager:0.2.0

Initial Release

machine-controller-manager declaratively manages machines/VMs across different cloud providers.

Notable features:

• Initial release supports management of machines, machine-sets, and machine-deployments
• Supported cloud providers: AWS, GCP, Azure, OpenStack
• Tries to drain machines before deletion
• Supports auto machine repair where machines with conditions such as KubeletNotReady, DiskPressure are replaced by the MachineSet controller
• Controller supports control-cluster (where CRDs and secrets reside) and target-cluster (where worker nodes are created)

Docker image: eu.gcr.io/gardener-project/gardener/machine-controller-manager:0.1.0