SUSE Virtualization (Harvester) v1.7.0 Release

SUSE Virtualization is a cloud native hyperconverged infrastructure platform solution optimized for running virtual machine and container workloads in the data center, multi-cloud and edge environments.

We are excited to announce the release of SUSE Virtualization (Harvester) v1.7! This release brings various new features, enhancements, and bug fixes to improve the overall user experience and stability.

This release upgrades the OS SUSE Linux Enterprise (SLE) Micro from 5.5 to 6.1, providing a more secure and resilient foundation. It also includes new important features such as:

• Nvidia MIG-based vGPUs
• Automatic VM Workload Rebalancing
• Hotplug NIC
• Pause-Ability of Node Upgrades
• Virtual Machine VLAN Trunking
• Volume snapshot in guest clusters

Feature Highlights of SUSE Virtualization v1.7

SUSE Virtualization v1.7 introduces several new features and enhancements to improve the user experience and system performance. Here are some highlights:

Harvester can now automatically detect GPUs that support Multi-Instance GPU (MIG)-based partitioning. This allows virtual machines to share vGPU devices backed by NVIDIA GPUs (such as the A100, H100, and H200) that offer this capability. To leverage this feature, you must define the MIG profiles for your GPU, enable the MIG configuration, and enable the vGPU device associated with the MIG configuration.

Demo Video: MIG-based vGPU Device Support

Starting with v1.7.0, Harvester supports external storage devices that use multipath I/O (MPIO) to provide redundant data paths while preventing device identification conflicts. Through the os.externalStorageConfig setting, you can define the precise configuration (including device denylisting and allowlisting) that the multipath daemon uses to manage devices and ensure storage stability.

Demo Video: Multipath Device Recognition and Management

The NIC hotplug/unplug feature allows you to add and remove network interface controllers (NICs) from running live-migratable virtual machines without downtime. When you hotplug a NIC, Harvester migrates the virtual machine and connects the specified interface through bridge binding.

Note: NIC hotplug of virtual machines created using the Harvester Node Driver and the Harvester Terraform Provider isn’t supported yet.

Demo Video: NIC Hotplugging and Hotunplugging

Harvester v1.7.0 introduces a new option to the upgrade-config setting that enables pausing of automatic node upgrades for all or specific cluster nodes. This capability is essential for scenarios requiring manual maintenance or verification on the nodes before the upgrade continues. Once these tasks are completed, you must then explicitly instruct Harvester to resume the upgrade process on the target nodes.

Demo Video: Pausable Node Upgrades

Harvester v1.7.0 allows you to create VLAN trunk networks with multiple, overlapping VLAN ID ranges. When a virtual machine is attached to a VLAN trunk network, the guest operating system and applications are allowed to send and receive packets tagged with any of the VLAN IDs within the specified range.

Demo Video: Virtual Machine VLAN Trunking

v0.1.25 of the Harvester CSI Driver supports volume snapshots, providing point-in-time snapshots and restoration for guest cluster workloads. You can create, list, and delete snapshots of persistent volumes, and create new volumes from existing snapshots. To use this feature, you must ensure that the CSI snapshot controller and the required manifests are deployed on the guest cluster.

Harvester CSI Driver v0.1.25 is used in the following RKE2 versions: v1.31.14+rke2r1, v1.32.10+rke2r1, v1.33.6+rke2r1, and v1.34.2+rke2r1.

Demo Video: Volume Snapshot in Guest Clusters

The Virtual Machine Auto Balance add-on allows Harvester to leverage the Kubernetes Descheduler for rebalancing of virtual machine workloads. When enabled on a multi-node cluster, the add-on deploys the Descheduler and a related configuration. The Descheduler evicts pods that are not optimally placed according to administrator-defined policies, enhancing resource utilization and improving cluster performance.

Demo Video: Automatic VM Workload Rebalancing

You can find more details about these new features and improvements in the SUSE Virtualization v1.7.0 Release Notes.

Harvester is built on a modern cloud-native architecture that leverages Kubernetes and various open-source projects (such as Longhorn, KubeVirt, Kube-OVN, and Containerized-Data Importer) based on RKE2. This architecture provides a scalable, flexible, and resilient platform for running virtual machine and container workloads.

Some of the key benefits of using Harvester include:

• 100% open source with a vibrant community
• Built on CNCF projects
• Fully integrated Rancher management
• Simplified operations with a user-friendly interface

To learn more reasons to use SUSE Virtualization for managing your virtualization and container workloads in a cloud-native way, please visit https://www.suse.com/products/rancher/virtualization/.