Skip to Content
Referenceoo CLI

oo backups

Nightly backups, daily backups and snapshots.

Nightly backups of this Sunport, and daily backups and snapshots of its machines, as Settings › Backups shows them. Nightly backups copy the Sunport's database, settings and files off its server every night. Daily backups and snapshots copy a machine's whole disk while it runs: the provider keeps the last seven daily backups, and a snapshot until you delete it. What you turn on here is charged to the prepaid balance, and counts toward the monthly server budget (server_budget_limit). An administrator restores an agent or service VM with oo backups restore.

Each command runs a job and follows it to its end: exit 0 done, 1 failed, 124 timed out while the job continues (oo jobs wait JOB_ID). The finished job's result is the backups as they stand. Reading needs recovery.read and changing recovery.manage, both with jobs.create. MACHINE is a machine's resource from oo backups list.

oo backups list

Check this Sunport's machines again, as Check again does, and print their backups as the job's result: nightly backups, then each machine (resource, kind, provider) with its daily backups and snapshots, and snapshots of machines that are gone. An agent VM on OpenOrange capacity lists its nightly copies (nightlyPoints, newest first: id nightly:TIME and createdAt, each its whole disk as it was that night, without its attached volumes), or why it has none yet (nightlyUnavailable). Prices and who asked show to administrators.

oo backups list [options]
ArgumentRequiredDefaultChoicesDescription
--no-waitnoprint the accepted job instead of following it
--timeout TIMEOUTno300.0
--interval INTERVALno2.0

It also takes the connection options.

oo backups nightly

Turn nightly off-server backups of this Sunport on or off. Turning them off deletes every one of them, which can't be undone: it asks first, or takes --yes.

oo backups nightly [options] state
ArgumentRequiredDefaultChoicesDescription
stateyeson, offon or off
--yesnoconfirm it now; an agent passes it only once the person confirmed it in chat
--no-waitnoprint the accepted job instead of following it
--timeout TIMEOUTno300.0
--interval INTERVALno2.0

It also takes the connection options.

oo backups daily

Turn a machine's daily backups on or off. Turning them off deletes the machine's backups now, which can't be undone: it asks first, or takes --yes. Daily backups OpenOrange set up can't be turned off (daily_backups_not_owner) until you turn them on here, which takes them over, charged from then on.

oo backups daily [options] MACHINE state
ArgumentRequiredDefaultChoicesDescription
MACHINEyesthe machine's resource, from oo backups list
stateyeson, offon or off
--yesnoconfirm it now; an agent passes it only once the person confirmed it in chat
--no-waitnoprint the accepted job instead of following it
--timeout TIMEOUTno300.0
--interval INTERVALno2.0

It also takes the connection options.

oo backups snapshot

Take a snapshot of a machine's whole disk. It is kept, and charged by its size each month, until you delete it; a machine keeps a limited number (snapshot_limit).

oo backups snapshot [options] MACHINE
ArgumentRequiredDefaultChoicesDescription
MACHINEyesthe machine's resource, from oo backups list
--name NAMEnoits name: letters, digits, spaces and . _ : ( ) / -
--no-waitnoprint the accepted job instead of following it
--timeout TIMEOUTno300.0
--interval INTERVALno2.0

It also takes the connection options.

oo backups delete-snapshot

Delete a snapshot taken here. It asks first, or takes --yes.

oo backups delete-snapshot [options] SNAPSHOT_ID
ArgumentRequiredDefaultChoicesDescription
SNAPSHOT_IDyesthe snapshot's ID, from oo backups list
--yesnoconfirm it now; an agent passes it only once the person confirmed it in chat
--no-waitnoprint the accepted job instead of following it
--timeout TIMEOUTno300.0
--interval INTERVALno2.0

It also takes the connection options.

oo backups restore

Put an agent or service VM back to one of its own daily backups or snapshots. This overwrites everything on the VM. A snapshot of the VM is taken first, which is the way back: it is charged until you delete it and counts toward the machine's limit (snapshot_limit: delete one first). The job follows the restore to success or failure. If the wait expires, refresh with oo backups list before retrying: the restore may still complete. An agent or service VM on Incus goes back to a snapshot an update took of it instead: IMAGE_ID is pre-update:JOB_ID, from preUpdateSnapshots in oo backups list. That overwrites its disk, and the VM restarts. Its current state is kept first as a snapshot for 7 days (kind restore), and restoring from that one undoes it. An agent's VM on OpenOrange capacity also goes back to one of its nightly copies: IMAGE_ID is nightly:TIME, from nightlyPoints in oo backups list, and VM_NAME is the agent's ID. Its whole disk goes back to that night: everything since is lost, attached volumes aren't included, and a secret deleted since comes back. Its current state is kept first as a safety snapshot, and the result's way_back undoes it (oo backups restore MACHINE pre-update:JOB_ID --name VM_NAME). The agent is then brought up to date, even while its updates are held, which stay held, and must answer a test message: the job succeeds only then. Otherwise it fails (nightly_restore_incomplete) saying what the re-seed did, what the test message got back and that way back; the same restore again within a day only brings the agent up to date. Only OpenClaw and Hermes agents, on a VM no other agent runs on (nightly_restore_shared_vm), for now. Only a person who is an administrator can ask; an agent's key is refused (administrator_required). It asks first, or takes --yes.

oo backups restore [options] MACHINE IMAGE_ID
ArgumentRequiredDefaultChoicesDescription
MACHINEyesthe machine's resource, from oo backups list
IMAGE_IDyesthe backup's or snapshot's ID, pre-update:JOB_ID for an update's or a restore's snapshot, or nightly:TIME for an agent VM's nightly copy, from oo backups list
--name VM_NAMEyesthe VM's name, to confirm which one is overwritten
--yesnoconfirm it now; an agent passes it only once the person confirmed it in chat
--no-waitnoprint the accepted job instead of following it
--timeout TIMEOUTno300.0
--interval INTERVALno2.0

It also takes the connection options.

On this page