Enhancements

This section outlines new features and enhancements introduced in the Mirantis Container Cloud release 2.17.0. For the list of enhancements in the Cluster releases 11.1.0 and 7.7.0 that are introduced by the Container Cloud release 2.17.0, see the Cluster releases (managed).


General availability for Ubuntu 20.04 on greenfield deployments

Implemented full support for Ubuntu 20.04 LTS (Focal Fossa) as the default host operating system that now installs on management, regional, and managed clusters for the following cloud providers: AWS, Azure, OpenStack, Equinix Metal with public or private networking, and non-MOSK-based bare metal.

For the vSphere and MOSK-based (managed) deployments, support for Ubuntu 20.04 will be announced in one of the following Container Cloud releases.

Note

The management or regional bare metal cluster dedicated for MOSK-based managed clusters is based on Ubuntu 20.04.

Caution

Upgrading from Ubuntu 18.04 to 20.04 on existing deployments is not supported.

Container Cloud on top of MOSK Victoria with Tungsten Fabric

Implemented the capability to deploy Container Cloud management, regional, and managed clusters on top of Mirantis OpenStack for Kubernetes (MOSK) Victoria that is based on the Tungsten Fabric networking.

Note

In the MOSK Victoria with Tungsten Fabric deployments of Container Cloud, Octavia enables a default security group for newly created load balancers. To change this configuration, refer to MOSK Operations Guide: Configure load balancing.

EBS instead of NVMe as persistent storage for AWS-based nodes

Replaced the Non-Volatile Memory Express (NVMe) drive type with the Amazon Elastic Block Store (EBS) one as the persistent storage requirement for AWS-based nodes. This change prevents cluster nodes from becoming unusable after instances are stopped and NVMe drives are erased.

Previously, the /var/lib/docker Docker data was located on local NVMe SSDs by default. Now, this data is located on the same EBS volume drive as the operating system.

Manager nodes deletion on all cluster types

TechPreview

Implemented the capability to delete manager nodes with the purpose of replacement or recovery. Consider the following precautions:

  • Create a new manager machine to replace the deleted one as soon as possible. This is necessary since after a machine removal, the cluster has limited capabilities to tolerate faults. Deletion of manager machines is intended only for replacement or recovery of failed nodes.

  • You can delete a manager machine only if your cluster has at least two manager machines in the Ready state.

  • Do not delete more than one manager machine at once to prevent cluster failure and data loss.

  • For MOSK-based clusters, after a manager machine deletion, proceed with additional manual steps described in Mirantis OpenStack for Kubernetes Operations Guide: Replace a failed controller node.

  • For the Equinix Metal and bare metal providers:

    • Ensure that the machine to delete is not a Ceph Monitor. If it is, migrate the Ceph Monitor to keep the odd number quorum of Ceph Monitors after the machine deletion. For details, see Migrate a Ceph Monitor before machine replacement.

    • If you delete a machine on the regional cluster, refer to the known issue 23853 to complete the deletion.

For the sake of HA, limited a managed cluster size to have only an odd number of manager machines. In an even-sized cluster, an additional machine remains in the Pending state until an extra manager machine is added.

Custom values for node labels

Extended the use of node labels for all supported cloud providers with the ability to set custom values. Especially from the MOSK standpoint, this feature makes it easy to schedule overrides for OpenStack services using API. For example, now you can set the node-type label to define the node purpose such as hpc-compute, compute-lvm, or storage-ssd in its value.

The list of allowed node labels is located in the Cluster object status providerStatus.releaseRef.current.allowedNodeLabels field. Before or after a machine deployment, add the required label from the allowed node labels list with the corresponding value to spec.providerSpec.value.nodeLabels in machine.yaml.

Note

Due to the known issue 23002, it is not possible to set a custom value for a predefined node label using the Container Cloud web UI. For a workaround, refer to the issue description.

Machine pools

Introduced the MachinePool custom resource. A machine pool is a template that allows managing a set of machines with the same provider spec as a single unit. You can create different sets of machine pools with required specs during machines creation on a new or existing cluster using the Create machine wizard in the Container Cloud web UI. You can assign or unassign machines from a pool, if required. You can also increase or decrease replicas count. In case of replicas count increasing, new machines will be added automatically.

Automatic propagation of Salesforce configuration to all clusters

Implemented the capability to enable automatic propagation of the Salesforce configuration of your management cluster to the related regional and managed clusters using the autoSyncSalesForceConfig=true flag added to the Cluster object of the management cluster. This option allows for automatic update and sync of the Salesforce settings on all your clusters after you update your management cluster configuration.

You can also set custom settings for regional and managed clusters that always override automatically propagated Salesforce values.

Note

The capability to enable this option using the Container Cloud web UI will be announced in one of the following releases.