Skip to content

apiary-pgsink

Replicate an Apiary SQLite database into PostgreSQL — backfill the history, then follow it — with per-table filters and injected fields.

Standalone by construction: its own repository, its own release cadence, and no Go dependency on Apiary in either direction. It reads the daemon's database; it is not a plugin and does not run inside the daemon.

brew install --cask orlandoburli/tap/pgsink

pgsink doctor      # check the table catalog against a live Apiary database
pgsink migrate     # create or alter the target tables
pgsink backfill    # load history
pgsink sync        # follow forever

Installation · Quickstart — five minutes from nothing to a PostgreSQL copy that stays current.

Backfill and sync are the same pipeline. Only the starting watermark and the stop condition differ, so the backfill path is exercised by every test the follower has.

Status

All four commands work, with Prometheus metrics, health checks and a poison-row quarantine. Not yet tagged for release.

Requirements

  • macOS or Linux, amd64 or arm64. A launchd agent and a systemd unit are both shipped in deploy/.
  • Runs on the same host as the Apiary daemon, as the daemon's user or in a group with write access to its data directory. SQLite in WAL mode has no true read-only reader — even a reader must be able to write the -shm wal-index file. See docs/deployment.md.
  • PostgreSQL 13 or newer as the target. Only this connection crosses the network.

Documentation

https://orlandoburli.com.br/apiary-pgsink

License

Licensed under the BSD 3-Clause License — the same terms as Apiary. No commercial license is required; see COMMERCIAL.md.