Extensions
If you are self-hosting, we do not recommend installing ParadeDB in your primary Postgres. ParadeDB is designed to run as a logical replica of your primary, which isolates search and analytical workloads from your primary.
If you already self-manage Postgres, you may prefer to install ParadeDB directly within your self-managed Postgres instead of deploying the ParadeDB Helm chart.
This can be done by installing the two extensions that ParadeDB uses:
- The
pg_search
extension powers all of ParadeDB’s full text search and analytics. - The
pg_analytics
extension powers ParadeDB’s integrations.
Prerequisites
Ensure that you have superuser access to the Postgres database.
Next, install libicu
.
Install the ParadeDB Postgres Extensions
ParadeDB Community
ParadeDB provides prebuilt binaries for our extensions on
- Debian 12
- Ubuntu 22.04 and 24.04
- Red Hat Enterprise Linux 8 and 9
- macOS 14 (Sonoma) and 15 (Sequoia)
Postgres 14, 15, 16, and 17 are available. If you are using a different version of Postgres or a different operating system, you will need to build the extensions from source.
pg_search
The prebuilt releases can be found in GitHub Releases.
You can replace 0.14.1
with the pg_search
version you wish to install and
17
with the version of Postgres you are using.
pg_analytics
The prebuilt releases can be found in GitHub Releases.
You can replace 0.3.0
with the pg_analytics
version you wish to install
and 17
with the version of Postgres you are using.
ParadeDB Enterprise
If you are a ParadeDB Enterprise user, you should have received a copy of the enterprise binaries. Please contact sales for access.
Update postgresql.conf
Next, add the extension(s) to shared_preload_libraries
in postgresql.conf
.
Reload the Postgres server for these changes to take effect.
Load the Extensions
Finally, connect to your Postgres database via your client of choice (e.g. psql
) and run the following command:
pg_search
can be combined with pgvector
for hybrid search. You can find
the instructions for installing pgvector
here.
Was this page helpful?