PgPaw is distributed as a single CLI binary named pgpaw.

Install with Cargo

Use Cargo when you want to build from source:

cargo install pgpaw

Install with npm

Use npm when a prebuilt binary is available for your platform:

npm install -g pgpaw

The npm package downloads the native binary during postinstall. If your platform is not supported by the prebuilt package, install with Cargo instead.

Verify the install

pgpaw --help
pgpaw init --help
pgpaw serve --help
pgpaw primary --help

The CLI should identify itself as pgpaw, and the default replication objects should be:

Runtime requirements

For normal pgpaw serve mode you need:

• Postgres 13 or newer;
• network access from PgPaw to upstream Postgres;
• logical replication enabled on upstream Postgres;
• a publication containing the tables PgPaw should serve;
• a persistent local --data-dir for the pglite replica.

For the one-time pgpaw init step, the Postgres user may need elevated permissions. It may run ALTER SYSTEM, create a publication, and install a DDL event trigger. If the user cannot install the DDL trigger, PgPaw still runs, but you should restart PgPaw after schema changes that affect served tables.

Files written by PgPaw

pgpaw serve writes the local pglite replica into --data-dir.

pgpaw serve --data-dir /var/lib/pgpaw ...

Use persistent storage in production. If the data directory is deleted, PgPaw must rebuild the local replica from upstream.