본문으로 건너뛰기
This page is not translated yet. 한국어 is switched on so the navigation and URLs can be tested; the text below is the English original. Use the language selector in the header to switch back, or see the documentation home.

Recover an unresponsive workstation

how-to · Reviewed 2026-07-30 · FEAT-0035 GA

If a workstation stops responding, you can restart it yourself from the UI. There are two options, reboot and reprovision, and the choice depends on how readily your coretype can be reacquired.

Neither loses data written to disk. Both kill running processes, so unsaved application state is lost.

How you'll find out​

Rescale checks connectivity to every node about once a minute. After 5 consecutive failures the workstation is flagged as unresponsive, and you get an hourly email while it stays that way.

Usual causes:

  • Your workload consumed all available resources, leaving the machine unable to respond.
  • Underlying hardware degradation on the host.
Charges continue while it's unresponsive

An unresponsive workstation is still billed. Recover or stop it: don't leave it sitting.

To turn the emails off: Preferences → Notifications, uncheck Workstation connectivity warning. Only the owner receives them.

Try waiting first​

If the cause was a temporary resource spike, the workstation may recover on its own as the workload progresses. If you know a heavy step is running, give it a few minutes before restarting and losing that work.

Reboot or reprovision​

RebootReprovision
What it doesShuts down and restartsSuspends and resumes
HardwareSame physical hostFresh hardware
Public IPPreservedChanges
Fixes host-level faultsNoYes
Time2–3 min (AWS) · 15–20 min (Azure)Similar to a resume
RequiresReboot support enabled; AWS or AzureA suspendable workstation

Which to pick​

Prefer reprovision when it's available. Moving to a new physical host clears hardware degradation, which a reboot can't.

Choose reboot if your coretype is capacity-constrained. Reprovisioning releases your hardware and asks for it back, and it may not be available. A reboot keeps the machine you have. This is the same capacity risk as suspending.

A reboot logs "Finished rebooting workstation" to the userlog when it's done.

Data preservation​

Survives
Data written to disk✅ Yes
Unsaved data in a running application❌ No
Attached HPS and CFS filesystems✅ Remounted automatically

Running simulations are killed by both operations, though their written output is kept.

Effect on collaborators​

Anyone connected over the display protocol is disconnected abruptly, without warning. Tell them first. See sharing.

If it's still broken​

Contact support@rescale.com. If the workstation failed rather than merely hung, the disk is retained for 7 days and data may be recoverable, but contact Support within that window. See saving and syncing files.

What isn't covered​

  • Batch jobs with desktop visualization don't support reboot or reprovision, and can't opt out of automatic termination on health-check failure.
  • Workstations Basic doesn't include reboot or reprovision.