Virtual machine actions

Edit online

If you have included Red Hat OpenShift Virtualization with a target configuration for Red Hat OpenShift, the following actions are supported for virtual machines:

Actions for virtual machines

  • Move

    Move a resource, such as a virtual machine (guest VM entity in Turbonomic) to another host to address performance issues or improve infrastructure efficiency. For example, if a particular host is congested for CPU, you can move the guest VM to a host with sufficient capacity. When a move is completed, the underlying resource entity in Red Hat OpenShift Virtualization is moved, not the workload controller that represents the resource.

    Turbonomic adheres to any affinity or anti-affinity rule configurations for guest VMs when it recommends move actions. For more information about affinity and anti-affinity rules, see Pod affinity and anti-affinity rules.

  • Resize (Up/Down)

    Resize vCPU and vMem allocations to ensure optimal utilization of resources. Resize actions are executed through the workload controller, which then rolls out the changes in the running environment.

    Resize actions support both scaling up to address performance issues and scaling down to improve efficiency and reduce costs.

    Turbonomic prevents resizing guest VMs below their minimum resource requirements. For example, if a guest VM is configured in Red Hat OpenShift Virtualization to run with a minimum of 2 VCPUs and 2 GiB of memory, Turbonomic does not resize the guest VM below these minimum requirements.

    Turbonomic does not resize guest VMs that are configured using instance types.

    For actions that are associated with Red Hat OpenShift Virtualization virtual machine pods, Turbonomic takes into consideration any run strategy that is defined in the virtual machine spec (RunStrategy field). For example, if a run strategy is defined so that a guest VM runs only once, Turbonomic does not recommend resize actions for that guest VM. For move actions, Turbonomic can change the run strategy to generate a disruptive resize action for the move. After the resize is completed, the run strategy is reset.

  • Attempt vCPU hot-plug

    When this setting is enabled, Turbonomic attempts to resize vCPU for guest VMs without restarting them. For more information, see Red Hat OpenShift Virtualization.

  • Attempt vMEM hot-plug

    When this setting is enabled, Turbonomic attempts to resize vMem for guest VMs without restarting them. For more information, see Red Hat OpenShift Virtualization.

The preceding actions are supported for Red Hat OpenShift Virtualization. For more information about the additional actions that are supported for an overall Red Hat OpenShift target, see Container platform actions.

Policy constraints and configuration for virtual machine actions

Actions for Red Hat OpenShift Virtualization virtual machines are controlled by multiple policy types. Understanding which policies affect each action type helps you configure the appropriate settings to control action generation and execution.

Move actions: Move actions for virtual machines are controlled by container pod policies. Since virtual machines in Red Hat OpenShift Virtualization run as pods, the default container pod policies determine whether move actions are generated and how they are executed.

  • Policy type: Container pod.

  • Default mode: Manual.

  • Configuration: Navigate to Settings > Policies > Automation policies and create or modify a container pod policy.

Move actions also respect pod affinity rules, anti-affinity rules, topology spread constraints, and security context constraints (SCCs). For more information, see Container pod move actions.

Resize actions: Resize actions for virtual machines (vCPU and vMem limit and request changes) are shaped by multiple policy types that work together to control both the generation and execution of these actions:

  1. Container spec policies: Control the generation of resize actions.

    • Define capacity ranges (min/max thresholds) for CPU and memory limits and requests.

    • Set increment constants for resize values.

    • Configure observation periods and aggressiveness settings.

    • Specify CPU throttling tolerance.

    • Configuration: Navigate to Settings > Policies > Automation policies and create or modify a container spec policy.

    For more information, see Container spec policies.

  2. Workload controller policies: Control the execution of resize actions.

    • Set action modes (Automatic, Manual, Recommend, or Disabled) for each resize action type.

    • Control whether resize up and resize down actions are automated or require manual approval.

    • Default mode: Manual for all resize actions (vCPU limit/request resize up/down, vMem limit/request resize up/down).

    • Configuration: Navigate to Settings > Policies > Automation policies and create or modify a workload controller policy.

    For more information, see Workload controller policies.

For a resize action to be generated and executed, the container spec policy must allow generation based on configured capacity ranges and constraints, and the workload controller policy must allow execution based on the configured action mode. If either policy prevents the action, it will not be generated or executed.