VM Operations

The virtbench vm-ops command group bundles day-2 VM lifecycle operations used during benchmarking and cluster validation. Each operation is a thin Click wrapper around a script in vm-ops/ that you can also invoke directly.

Use Case: Drain nodes, rebalance VMs across hosts, snapshot VMs in batches, run blkdiscard inside guests, and power VMs on or off — all without leaving the unified CLI.

Available Operations

Operation	Purpose
`drain-nodes`	Drain Kubernetes nodes and measure drain time
`rebalance-vms`	Rebalance VMs evenly across worker nodes
`vm-snapshot`	Create `VirtualMachineSnapshots` in batches
`run-blkdiscard`	Run `blkdiscard` on data disks inside VMs
`power-toggle-vms`	Power VMs on or off (`--action {on,off}`)

Quick Reference

# Drain two nodes in parallel
virtbench vm-ops drain-nodes --nodes worker-1 worker-2 --parallel

# Rebalance the cluster (dry-run)
virtbench vm-ops rebalance-vms --vm-name rhel-elbencho-1 --dry-run

# Snapshot 50 VMs in batches of 25, 5 minutes apart
virtbench vm-ops vm-snapshot \
  --namespace-prefix migration --start 1 --end 50 \
  --vm-name rhel-9-vm --batch-size 25 --interval 300

# Run blkdiscard on auto-detected data disks
virtbench vm-ops run-blkdiscard \
  --namespace-prefix rhel-eb-filler --start 1 --end 10 \
  --vm-name rhel-elbencho-1

# Power off 50% of running VMs on a node
virtbench vm-ops power-toggle-vms \
  --action off --node worker-1 --percentage 50

# Power them back on from the saved list
virtbench vm-ops power-toggle-vms \
  --action on --vm-list-file powered_off_vms_worker-1_<ts>.txt

Common Patterns

Discovery Modes

Most operations support two ways of selecting target VMs:

Range mode — --namespace-prefix <prefix> --start <i> --end <j> plus --vm-name <name>. The CLI iterates prefix-i through prefix-j and acts on the named VM in each namespace.
Node mode — --node <node> discovers VMs running on a given worker. Only available where the running VMI is bound to a node (for example, power-toggle-vms --action off).

Concurrency

All long-running operations expose --concurrency (a few use -c as shorthand) so you can dial parallelism up or down per operation. Sensible defaults are wired into each script.

Dry-Run

Every operation supports --dry-run. Use it to preview the work without mutating cluster state.

Logging

drain-nodes, vm-snapshot, and run-blkdiscard accept --log-file <path> and the global virtbench --log-file <path>. When neither is provided, a timestamped log file is generated automatically.

rebalance-vms and power-toggle-vms print to stdout/stderr and do not write a log file.

Prerequisites

kubectl (or oc) on $PATH, with a working kubeconfig.
virtctl (or the kubectl virt Krew plugin) for VM power operations.
For run-blkdiscard: an SSH-capable pod in the cluster (default ssh-test-pod in the default namespace) with network access to the target VMs.