Run, back up and recover

Run as a foreground process

nookins service foreground

Keep that terminal open. Use another terminal with the same NOOKINS_HOME for the console and diagnostics. Stop foreground mode with Ctrl-C before offline configuration or switching to the user service.

Install the Linux user service

nookins service install
nookins service start
nookins service status
nookins service logs --lines 100

Use the service reference for options and prerequisites. Do not run a second foreground supervisor over the same home. To stop or restart a managed service:

nookins service stop
nookins service restart

The operator console can inspect jobs, approvals, schedules and usage. Operations commands distinguish inspecting work from cancelling or changing it. Check the exact task and generation before cancellation.

Create and verify a private backup

Stop Nookins before taking a maintenance snapshot:

nookins service stop
nookins maintenance backup create nookins-backup.tar.gz
nookins maintenance backup verify nookins-backup.tar.gz

A snapshot contains private configuration, keys, authentication, conversations and other durable data. Protect it like the installation itself; never attach it to public feedback. Large re-downloadable assets are excluded unless --include-assets is selected. Verify before relying on a backup.

Restore deliberately

Read nookins maintenance backup restore --help and the backup reference for the exact destination and options. Restoration requires an absent target and is atomic, not a merge into a running home. A backup does not authorize cloning an ownership identity or converting Ward state into Nookins. Copied/remounted identities and cross-product restores are outside the public fresh-home qualification.

Preserve the original installation and the verified snapshot. On refusal, retain the error and reported recovery records instead of deleting ownership or state files. Stop the new instance before returning to the original; never let both answer the same messaging account. Restore does not reverse messages, purchases or other external effects.

Update safely

Follow updates for signed feed checks and the maintenance layout. Keep interrupted-operation records and use the exact recovery path reported by the command. Native updates preserve the existing home; do not recreate agents after a normal update.

Diagnose the first failing layer

Check CLI diagnostics, service readiness, provider status, account connection, agent grants and route identity in that order. Record the version, exact command, exit/error text and a redacted operation ID. Troubleshooting explains what to share.