Enhancements

This section outlines new features implemented in the Cluster release 17.1.0 that is introduced in the Container Cloud release 2.26.0.

Support for MKE 3.7.5 and MCR 23.0.9

Introduced support for Mirantis Container Runtime (MCR) 23.0.9 and Mirantis Kubernetes Engine (MKE) 3.7.5 that supports Kubernetes 1.27.

On existing MOSK clusters, MKE and MCR are updated to the latest supported version when you update your cluster to the Cluster release 17.1.0.

Support for Rook v1.12 in Ceph

Added support for Rook v1.12 that contains the Ceph CSI plugin 3.9.x and introduces automated recovery of RBD (RWO) volumes from a failed node onto a new one, ensuring uninterrupted operations.

For a complete list of features introduced in the new Rook version, refer to official Rook documentation.

Support for custom device classes in a Ceph cluster

TechPreview

Implemented the customDeviceClasses parameter that enables you to specify the custom names different from the default ones, which include ssd, hdd, and nvme, and use them in nodes and pools definitions.

Using this parameter, you can, for example, separate storage of large snapshots without touching the rest of Ceph cluster storage.

Network policies for Rook Ceph daemons

To enhance network security, added NetworkPolicy objects for all types of Ceph daemons. These policies allow only specified ports to be used by the corresponding Ceph daemon pods.

Upgraded logging pipeline in StackLight

Completely reorganized and significantly improved the StackLight logging pipeline by implementing the following changes:

  • Switched to the storage-based log retention strategy that optimizes storage utilization and ensures effective data retention. This approach ensures that storage resources are efficiently allocated based on the importance and volume of different data types. The logging index management implies the following advantages:

    • Storage-based rollover mechanism

    • Consistent shard allocation

    • Minimal size of cluster state

    • Storage compression

    • No filter by logging level (filtering by tag is still available)

    • Control over disk space to be taken by indices types:

      • Logs

      • OpenStack notifications

      • Kubernetes events

  • Introduced new system and audit indices that are managed by OpenSearch data streams. It is a convenient way to manage insert-only pipelines such as log message collection.

  • Introduced the OpenSearchStorageUsageCritical and OpenSearchStorageUsageMajor alerts to monitor OpenSearch used and free space from the file system perspective.

  • Introduced the following parameters:

    • persistentVolumeUsableStorageSizeGB to define exclusive OpenSearch node usage

    • output_kind to define the type of logs to be forwarded to external outputs

Important

Changes in the StackLight logging pipeline require the following actions before and after the manged cluster update:

Support for custom labels during alert injection

Added the alertsCommonLabels parameter for Prometheus server that defines the list of custom labels to be injected to firing alerts while they are sent to Alertmanager.

Caution

When new labels are injected, Prometheus sends alert updates with a new set of labels, which can potentially cause Alertmanager to have duplicated alerts for a short period of time if the cluster currently has firing alerts.