Managing virtual machines in RLSI

Managing virtual machines in RLSI

The VMS tab in the SC//Reliant™ Local Status Indicator (RLSI) lists the virtual machines (VMs) on the node you are connected to and holds their power controls. The BACKUPS tab backs them up and restores them.

IMPORTANT: Every action on these tabs applies to the node you are connected to. Confirm the node and the cluster state before using any of them. See Before you act.

Viewing your virtual machines

The VMS tab lists every VM on the node.

ColumnDescription
IdThe internal ID assigned to a running VM.
VMThe VM name.
StatusThe VM's current state, such as running.
ActionsThe power controls for that VM.

Buttons that do not apply to a VM's current state are greyed out — Start is unavailable on a VM that is already running, for example.

The Status column refreshes on an interval rather than instantly, so it can be up to 30 seconds behind what the VM is actually doing. If a status looks wrong, wait and refresh before concluding an action failed.

Rebooting a virtual machine

Reboot is the standard fix for most VM issues.

  1. On the VMS tab, find the VM and check its Status is running.

  2. Select Reboot in that VM's row.

  3. Confirm the action. Your browser asks you to confirm and names the VM — read the name before selecting OK, as this is the check that catches an action aimed at the wrong row.

  4. Watch the Output log below the list.

    Rebooting kvm_cluster_50...
Reboot successfully triggered.

    Reboot successfully triggered means the guest operating system accepted the restart signal. It does not mean the VM has finished restarting.

  5. Wait for the guest to start, then check the Status column shows running again. A VM can be running before the application inside it is ready, so where it matters, confirm the application is reachable.

The available actions

ActionWhat it doesSafe to use?
StartStarts a stopped VM, or resumes a paused one.Yes.
ShutdownShuts down a running VM cleanly, as if shut down from inside the guest.Yes.
RebootShuts down and restarts a running VM cleanly.Yes.
Hard ResetResets the VM immediately, like pressing a reset button.Caution. Data loss is possible.
Hard ShutdownCuts power to the VM, like pulling the plug.Caution. Data loss is possible.

Always try the graceful action first. Shutdown and Reboot ask the guest operating system to do the work, so applications close cleanly. Hard Reset and Hard Shutdown bypass the guest, risking unsaved data and leaving the filesystem needing repair.

A graceful action depends on the guest responding. If the guest has hung it will be ignored and the VM stays up — that is what the hard actions are for. If you are unsure, escalate rather than forcing.

Backing up a virtual machine

IMPORTANT: A full backup shuts the VM down for the whole operation, so the workload inside it is unavailable. Treat it as planned maintenance rather than a background task. RLSI also puts the cluster into maintenance mode first, so that the clustering software does not treat the shutdown as a failure.

Use Full Backup rather than Snapshot. A full backup is a complete, independent file that can be restored on its own; a snapshot is only a local record of changes against a disk on the same node, so it is not a recovery point if that node is lost.

  1. Open the BACKUPS tab and select Manual Backup.

  2. In SELECT DOMAIN, choose the VM.

  3. For TYPE, select Full Backup.

  4. Select the folder under LOCATION.

  5. For OUTPUT FORMAT, choose Compressed. It produces a smaller .tar.gz file and takes a little longer than Standard, which produces an uncompressed .tar.

    The Restore Backup screen lists .tar.gz among the file types it accepts but does not list .tar, so Compressed is the safer choice for any backup you might need to restore.

  6. Accept or edit the FILE NAME. The name is filled in for you and includes the VM name and a timestamp, so accepting it avoids overwriting an earlier backup. Only letters, numbers, dashes, and underscores are allowed.

  7. Select Backup now. A banner confirms the operation started and the form's controls are disabled until it finishes.

  8. Watch the Operation Logs below the form.

    Starting full backup for kvm_cluster_50
Maintenance mode set to enabled.
Shutting down kvm_cluster_50...
checking status 30...

    checking status counting downward is a wait loop, not an error — RLSI is polling the VM while it shuts down. If the countdown runs all the way out, the VM did not stop, so capture the log and escalate.

  9. When the operation finishes, confirm on the VMS tab that the VM is running again, and confirm in the header bar that the cluster is no longer in maintenance mode.

    Both matter. A VM left shut down is an outage, and a cluster left in maintenance mode is not protecting its workloads.

Restoring a virtual machine

IMPORTANT: A restore replaces the VM's current disk. The VM is unavailable while the restore runs, and anything written since the backup was taken is lost. Confirm you have the right VM and the right backup before you start.

  1. Open the BACKUPS tab and select Restore Backup.

  2. In SELECT DOMAIN, choose the VM.

  3. Under RESTORE FROM, choose where the backup is coming from.

    OptionUse it for
    Automatic backupsScheduled backups, which restore from differential snapshots. If none are configured for this VM, RLSI reports There are no automatic backups to restore from.
    Specific fileA backup file in the node's backup folder, such as one you took yourself from Manual Backup.

  4. If you chose Specific file, the contents of the backup folder appear under LOCATION. Select the file you want.

    The listing shows everything in the folder rather than only backups, so check you have selected a backup file. Valid files have a .qcow2 or .tar.gz extension.

    Restoring a "Headless" .qcow2 image may fail, according to the warning on the form itself.

  5. Select Restore backup.

  6. Watch the Operation Logs until the restore reports completion.

  7. Confirm on the VMS tab that the VM is running, and confirm in the header bar that the cluster is not left in maintenance mode.

© 2026 Scale Computing, Inc. All rights reserved.

Scale Computing, SC//AcuVigil, SC//Connect, and SC//Reliant are trademarks of Scale Computing, Inc.