`maand snapshot`

Catalog database snapshots for workers labeled maand.

`maand snapshot backup`

Creates a consistent copy of data/maand.db (SQLite online backup) and rsyncs it to every worker with the maand label:

/opt/maand/<bucket_id>/backups/maand-<update_seq>.db

Example filename: maand-42.db. Re-running backup at the same update_seq overwrites that file.

Then keeps only the N highest update_seq backups on those workers according to backup_retention_count in maand.conf (default 7; 0 keeps forever). Legacy timestamp-named maand-*.db files are removed when pruning is enabled.

Requirements

Workers you want to receive backups must include label maand in workspace/workers.json (then maand build).
SSH/rsync reachability same as deploy.

Behavior

Aspect	Detail
Targets	`data.GetWorkers(..., ["maand"])`
Local snapshot	Under `tmp/`, then transferred
Failure	Non-zero exit (unlike deploy’s best-effort post-commit backup)
No targets	Exits 0 and logs that nothing was targeted

Deploy also runs this path after commit + VACUUM (best-effort). Use the CLI when you want a backup without a full deploy (uses the current catalog update_seq; does not bump it).

Restore material in the catalog: each maand build copies onto the singleton bucket row:

Disk	Column
`maand.conf`	`maand_conf`
`secrets/kv.key`	`kv_key`
`secrets/ca.key`	`ca_key`
`secrets/ca.crt`	`ca_crt`
`workspace/workers.json`	`workers_json`
`workspace/bucket.conf`	`bucket_conf`
`workspace/disabled.json`	`disabled_json`
`workspace/bucket.jobs*.conf`	`bucket_jobs_confs` (JSON map)
`.ssh/known_hosts`	`known_hosts`

Job trees live in job_files (already in the catalog). Catalog backups on maand-labeled workers therefore contain CA/KV secrets, workspace sources, and SSH host pins — treat those hosts like the CLI host.

`maand snapshot list`

Lists remote catalog backups on every worker labeled maand:

Column	Meaning
Worker	Worker IP
Update Seq	Parsed from `maand-<update_seq>.db`
Name	Backup filename
Path	Full path under `/opt/maand/<bucket_id>/backups/`

Sorted highest update_seq first. Rows are per worker (same filename on two hosts appears twice). With no maand workers, exits 0 and prints an empty table (and a short log line).

`maand snapshot restore`

# Copy a backup off a worker, then restore offline:
#   scp agent@worker:/opt/maand/<bucket_id>/backups/maand-42.db .
maand snapshot restore ./maand-42.db
maand snapshot restore ./maand-42.db --output /path/to/new-bucket

Reads a local catalog DB file (no SSH, no existing local bucket required) and writes a complete bucket tree under --output (default .):

Copies the file to data/maand.db
Writes maand.conf, secrets/kv.key, secrets/ca.key, secrets/ca.crt
Writes secrets/ca-trust.crt from KV maand/worker / certs/ca-trust.crt (falls back to ca.crt if missing)
Writes .ssh/known_hosts from the bucket row (empty file if none was stored)
Recreates workspace/ from bucket columns (workers.json, bucket.conf, disabled.json, bucket.jobs*.conf) and exports job_files verbatim into workspace/jobs/
Scaffolds templates/, logs/, tmp/, and .gitignore

Fails if output/data/maand.db already exists. Does not restore secrets/<ssh_key> — copy the SSH private key before deploy.