Thora Cloud
A direct line to Thora

Start a conversation

What can we help make simpler?

Tell us what you need. A real person will point you to the right place.

Email Thorahello@thora.cloudExplore SupportSee the available support paths
Browse docs Check virtual machines

Help and guides

Check virtual machines

View your virtual machines and see which operations need attention.

Before you start

Choose the organization that owns the virtual machine. Managed VM creation may be unavailable during beta.

Steps

  1. Open Compute and find Virtual machines.
  2. Choose Refresh to read the latest state.
  3. Open the machine you want to inspect.
  4. Read any Needs attention message and review the exact operation before approving it.

What to expect

You can see the machine state and pending operations. A submitted operation is complete only after its status confirms completion.

If something goes wrong

If Create managed VM is unavailable, the dashboard is not offering creation for your account. Do not assume high availability or automatic retries; use the availability and recovery information shown for the resource.

Technical reference

Read the resource before acting

Compute responses expose a customer resource ID, name, purpose, capacity, recorded lifecycle state, observation time, and operation availability. Provider host, runtime ID, and provider operation IDs stay private.

  • Unknown is not stopped.
  • Provisioning entries are reservations or jobs, not ready VMs.
  • Capacity, billing, high availability, SSH, diagnostics, and maintenance availability are explicit response fields.
  • Never select a provider host, runtime ID, image, public address, network, SSH key, user data, or SKU unless a future customer contract explicitly exposes it.

Lifecycle requests are proposals

Start, stop, restart, fixed-SKU provision, snapshot, restore, destroy, and snapshot removal are exact actions in the management model. The current customer MCP surface proposes start, stop, restart, provision, snapshot, and restore only; it deliberately excludes VM destruction and snapshot removal. A signed-in direct administrator reviews every exposed proposal's exact target and digest.

  • Proposals expire after a bounded review window.
  • Reuse the same idempotency key after an uncertain response.
  • Restore creates a new generation rather than overwriting an existing VM.
  • Destroy releases capacity only after verified deletion.
  • Catalog removal can precede physical encrypted-byte cleanup.

Use fixed read-only diagnostics

  • Health summary: guest OS state, failed-unit count, uptime, memory, and root-disk capacity.
  • Failed services: at most 20 failed systemd service names and bounded states, without journal content.
  • Network summary: sanitized interfaces and addresses, without MACs, routes, DNS, sockets, ports, or traffic statistics.
  • No diagnostic accepts a shell command, service selector, destination, path, file, or log source.

Know what a snapshot proves

A disk snapshot does not save RAM and does not by itself prove application or database consistency. Complete recovery evidence verifies the restored generation, application data, dependency state, and public service separately.