Power Loss Recovery

Power-loss recovery

What this page will help you do

Understand what happens when the power goes out during a print, and how to continue the print instead of starting over.

You do not have to set anything up. NebulaOS does this by itself.

What you need

Nothing. Power-loss recovery is part of NebulaOS and is always running during a print.

How it works, briefly

While a print runs, NebulaOS keeps saving a checkpoint of where the print is - the position in the file, the nozzle and bed temperatures, the bed mesh in use, and which objects you have cancelled if you are using Cancel Object.

The checkpoint is stored in two places: a tiny record on a memory chip on the printer's board that survives losing power, and a larger file on the printer's storage. The chip record is written last, and only once the file is safely on disk, so the printer never trusts a checkpoint that did not completely land. That ordering is the whole reason a power cut at the wrong moment does not leave you with a corrupt recovery.

You do not need to know any of this to use it. It matters only because it explains why the printer sometimes says there is nothing to recover: if the checkpoint did not fully land, it is not offered, rather than being offered and then failing.

Step-by-step: after the power comes back

1. Switch the printer on as usual.
2. Wait for it to start up - about a minute.
3. If a recoverable print was interrupted, the screen offers it to you, naming the file.
4. Choose:
   • Resume - carry on from the checkpoint.
   • Discard - forget the checkpoint and start fresh.

Look at the part before you choose. Resuming only makes sense if the print is still stuck to the bed in the position it was in. If it came loose, or you have moved the bed or the part, discard it.

No screenshot yet: the recovery offer appears as a prompt during startup, which the simulator's screenshot tour does not reach. This one needs a photograph of a real printer after a power cut.

Expected result

After choosing Resume, the printer works through a deliberate order:

1. The bed heats first, with the nozzle off.
2. Z lifts and X and Y home, so the toolhead is away from your part.
3. Only then does the nozzle heat, at the park position.
4. The bed mesh from the interrupted print is reloaded, the printer returns to the position it had reached, and printing continues from the same point in the file.

Objects you had cancelled stay cancelled, and if you were part-way through one, that object is started again rather than skipped.

A timelapse keeps its earlier frames. If you were recording a timelapse, the frames taken before the power cut are put back in front of the new ones when the print ends, so the finished video covers the whole print instead of only the part after the resume. Nothing to switch on - it is set up in the shipped configuration (machine.cfg, timelapse_frame_path).

Why the order matters. It would be simpler to heat everything at once, and a power-cut test on a real printer showed why that is wrong: the nozzle heated while still sitting over the part, which risks melting into the print you are trying to save. Heating the nozzle only once it has moved away costs a little time and removes that risk entirely.

The order comes from extensions f498796, the version NebulaOS 1.0.0 ships.

If a resume fails partway

Worth knowing, because the printer's behaviour here is deliberate and might otherwise look like a malfunction.

If a resume starts but cannot complete, NebulaOS cleans up rather than leaving things half done:

• The nozzle heater is switched off, so a hot nozzle is not left parked over your part.
• The file is unloaded, so the screen and Mainsail do not show a print that is not actually running.
• The bed heater stays on. This is on purpose: a warm bed keeps the part stuck down, which is exactly what you need in order to try again.
• The checkpoint is kept, so Resume is offered again.

So a failed resume leaves you able to retry, with the part still attached. If you would rather not retry, choose Discard and turn the bed off yourself.

From extensions f498796, the 1.0.0 release pin (nebulaos_power_loss_recovery.py, _undo_failed_resume).

Troubleshooting

What happens	What it means
No recovery is offered after a power cut	No complete checkpoint landed, so there is nothing to recover. Start the print again.
Recovery is offered but the part came loose	Choose Discard. Resuming onto a moved part will fail, or drive the nozzle into it.
Resume heats up, then stops	A step was refused. The nozzle is turned off, the bed left on, and Resume is offered again.
The print resumes in the wrong place	Report it with the logs: Settings → System → Report a problem.
You want the offer to go away	Choose Discard.

For advanced users: the commands behind it

The screen uses macros that call these. You can also run them from the console (Home → Console) or from Mainsail:

Command	What it does
NEBULAOS_PLR_STATUS	Reports the current power-loss-recovery status
NEBULAOS_PLR_RESUME	Resumes from the checkpoint
NEBULAOS_PLR_DISCARD	Discards any checkpoint

Two further commands exist, _NEBULAOS_PLR_LOAD_MESH and _NEBULAOS_PLR_SET_Z. The leading underscore marks them internal: they are steps of NEBULAOS_PLR_RESUME and are not meant to be run on their own.

Recovery and reversal

Discard is how you undo a pending offer, and it cannot be undone itself - the checkpoint is gone afterwards. It does not touch the G-code file, so you can always start the print again from the beginning.

Related

• Printing - pausing, resuming and cancelling a print normally
• Recovery - recovering the printer itself, which is a different thing