Skip to content

Set up Fotobank for someone else

Hand over a library the person can browse, back up, and recover. Use this procedure when an agent or administrator prepares their first library. Fotobank is pre-alpha. Start with copies of a small collection, keep independent originals, and do not promise that a successful trial makes it safe to discard them.

Agree on the setup

Ask the person to choose:

  • The machine and OS account that will run Fotobank. Application commands run there, under that account; an import path is not a remote upload.
  • Separate folders for the catalog, original files, and thumbnails. A local disk is enough for a trial. Confirm available space: imports copy files, checkouts make further copies, and backups need separate capacity.
  • A source folder containing sample copies and a separate backup destination.
  • Who may access the web app. Start with loopback access. Do not expose the default single-user identity to a network as though it were a login system.
  • Whether any photos may be sent to an AI provider. Leave AI disabled unless the person explicitly approves the provider and processing policy.

Do not choose a storage path simply because a missing mount left an empty directory. Do not change repository visibility, publish a share, enable external processing, or delete source files as part of setup.

Install and initialize

Follow setup for the supported installation path and folder creation. Use one explicit configuration selection throughout the session:

export FOTOBANK_CONFIG="$HOME/.config/fotobank-trial/config.toml"
fotobank config init

These are POSIX-shell examples. In PowerShell, set the same environment variable with $env:FOTOBANK_CONFIG = 'C:\path\to\config.toml'. If a configuration already exists, inspect it rather than replacing it. Use absolute storage paths so a different working directory cannot select a different library. Keep configuration on available local storage outside the photo folders; its directory must be writable for daemon runtime files.

Edit the three storage roots as described in setup, then inspect them:

fotobank config validate
fotobank config diagnose --json

Before the first start, missing catalog and uninitialized vault reports are expected. Missing storage directories or an unavailable mount are not. For an existing library, stop and investigate missing data instead of initializing a replacement.

Prove the first import

fotobank daemon start
fotobank config diagnose --json
fotobank import /path/to/sample-copies --json
fotobank media list --limit 5 --json

Record the web URL printed by startup. Check exit codes and the import result's failures, not just whether some photos appeared. Open the web app and inspect several photos. A missing thumbnail may be a format limitation, not a missing original.

Select an ID from the returned items and verify an independent download:

fotobank media show <media-id> --json
fotobank media download <media-id> --output /existing-folder/check.jpg --json

Use a new destination filename and the appropriate extension. Compare the returned SHA-256 with the input file's checksum. Check a related RAW or XMP attachment too if the sample collection contains one. The download command verifies bytes against the catalog; comparing with the source additionally checks that you imported the intended file.

Prove recovery before a larger import

Create a small album before taking the backup so the drill checks catalog choices as well as stored files:

fotobank albums create "Recovery check" --json
fotobank albums add <album-id> <media-id> --json
fotobank albums show <album-id> --json

Use the album id returned by create and a photo ID from the first import. Save the album ID, its member IDs, and the source checksums for comparison after restoration.

Follow backup and restore: initialize a separate repository, create and verify an archive, then restore into separate storage. Confirm the saved album membership and compare downloaded originals with the saved checksums in the recovered library. Do not delete the original deployment to simulate a loss. Keep optional AI and scheduled backups disabled in the recovered test copy.

Only enable scheduled backups after the person approves their destination and retention count. Repositories are not encrypted. Configuration and provider credentials need separate protection because archives do not contain them.

Hand over an operable library

Give the person the web URL, installed version, OS account, configuration path, storage paths, backup location and schedule, and the result of the recovery drill. Do not include credentials in the handoff or logs. Include:

fotobank daemon status --json
fotobank daemon stop
fotobank daemon start

Explain whether a supervisor or Fotobank's own background launcher starts the process. For a supervised installation, use the supervisor to stop or restart it. Point to automation for JSON results and retry rules, checkouts for editing, and optional AI before enabling providers. A running daemon is not proof that backups or queued thumbnails have finished; verify those results separately.