Rotom (wip)

Optional, and worth turning on, it makes the scanner coming back the success signal.

Set ROTOM_ENABLED=true, ROTOM_URL (RotomNG's HTTP listener, usually :7072) and ROTOM_SECRET, and Rotom becomes part of the update itself rather than a status widget.

What changes

Before an install

The box is disabled in Rotom, so the controller stops handing it accounts instead of a scan dying mid-session.

After it succeeds

The box is enabled and the scanner restarted onto the new build, replacing the post-install hook you would otherwise write per group.

The success signal

The box reappearing in Rotom with live workers, not `pm` having printed Success. If it does not come back within five minutes, the job log says so.

That third point is the reason to turn this on. An install can succeed and the scanner still fail to come up, a bad build, a wiped login, a hook that raced. Without Rotom, Magnemite calls that a success and moves to the next device. With it, the rollout stops.

Crash safety

If the hub restarts mid-install, the next scheduler tick re-enables any box it had disabled for a job that is no longer running. Nothing is left parked out of the pool because a Node process went away.

Matching devices

Devices are matched to Rotom by origin, falling back to the serial, and then to an unambiguous public IP. Once at least one box is matched:

the fleet table grows a Scanner column,
the device page gains a Restart scanner button.

A box that never matches keeps working normally, it simply uses its group's hooks instead.

Configuration

.env

ROTOM_ENABLED=true
# /api is appended, so either form works.
ROTOM_URL=http://rotom:7072
# Matches `secret` under http_listener in the Rotom config.
ROTOM_SECRET=…

Built against the RotomNG API.

GET /api/device, PUT /api/device/{id}/action/{action}, authenticated with X-Rotom-Secret.

If Rotom runs in another compose project, the hub needs to reach it, either put both on a shared external network, or point ROTOM_URL at its published host and port.