Place a VM on an eligible cluster

A reachable cluster is not automatically an allowed destination. Shepherd separates what a cluster can do from what policy permits, then uses that context when a reviewer chooses where a requested VM may run.

Browser demo · No account · Simulated data

Prepare infrastructure and business context

A connected installation needs suitable cluster access and registered namespaces. Systems and Services express business ownership; they are not a fixed mapping to one cluster. Maintain the catalogs used by requests before offering a deployment target.

Distinguish capability from allowance

Detected KubeVirt features and storage classes describe objective capabilities. Cluster policy supplies administrative allowance. A cluster can support a feature while policy prevents using it for a particular workload.

Keep concrete placement on the review side

Requesters see catalogs and sanitized placement hints. Reviewers work with concrete cluster and storage decisions. Check environment, workload requirements and policy before execution, then confirm the resulting VM on the selected destination.

Walk through the task

  1. Prepare a connected test cluster

    Verify Kubernetes/KubeVirt connectivity and the required permissions. Follow the deployment or chart instructions for managed-cluster access.

    Read deployment requirements
  2. Check namespace and workload context

    Review the registered namespace environment, requested template and instance size. These inputs drive candidate evaluation.

  3. Inspect capability and policy separately

    Resolve missing cluster features independently from an explicit administrative denial. A healthy connection alone does not make a destination eligible.

    Read the placement enforcement order
  4. Confirm placement and verify execution

    During review, choose an eligible cluster and storage configuration without changing the submitted namespace. Inspect the delivered resource and any provider errors.

    Read reviewer-controlled placement

Requirements and limits

  • Multi-cluster placement does not imply automatic cross-cluster migration, disaster recovery or application failover.
  • Placement guidance is not a capacity reservation or a provisioning guarantee.
  • The demo cannot verify cluster credentials, live storage or infrastructure policy enforcement.

Documentation and implementation

Common questions

Why is a healthy cluster ineligible?

Connectivity is only one input. Namespace environment, catalog requirements, capabilities and policy can rule out a destination.

Does the requester choose the cluster?

No. The create payload omits cluster_id. The reviewer selects a target using the submitted workload and placement context.