Skip to content

Using PostgreSQL

DETAILS: Tier: Free, Premium, Ultimate Offering: GitLab.com, Self-managed, GitLab Dedicated

As many applications depend on PostgreSQL as their database, you eventually need it in order for your tests to run. Below you are guided how to do this with the Docker and Shell executors of GitLab Runner.

Use PostgreSQL with the Docker executor

To pass variables set in the GitLab UI to service containers, you must define the variables. You must define your variables as either Group or Project, then call the variables in your job as shown in the following workaround.

In Postgres 15.4 and later, Postgres does not substitute a schema or owner name into an extension script if the name contains a quote, backslash, or dollar sign. If the CI variables are not configured, the value uses the environment variable name as a string instead. For example, POSTGRES_USER: $USER results in the POSTGRES_USER variable being set to '$USER', which causes Postgres to show the following error:

Fatal: invalid character in extension

The workaround is to set your variables in GitLab CI/CD variables or set variables in string form:

  1. Set Postgres variables in GitLab. Variables set in the GitLab UI are not passed down to the service containers.

  2. In the .gitlab-ci.yml file, specify a Postgres image:

    default:
       services:
         - postgres
  3. In the .gitlab-ci.yml file, add your defined variables:

      variables:
        POSTGRES_DB: $POSTGRES_DB
        POSTGRES_USER: $POSTGRES_USER
        POSTGRES_PASSWORD: $POSTGRES_PASSWORD
        POSTGRES_HOST_AUTH_METHOD: trust

    For more information about using postgres for the Host, see How services are linked to the job.

  4. Configure your application to use the database, for example:

    Host: postgres
    User: $POSTGRES_USER
    Password: $POSTGRES_PASSWORD
    Database: $POSTGRES_DB

Alternatively, you can set variables as a string in the .gitlab-ci.yml file:

variables:
  POSTGRES_DB: DB_name
  POSTGRES_USER: username
  POSTGRES_PASSWORD: password
  POSTGRES_HOST_AUTH_METHOD: trust

You can use any other Docker image available on Docker Hub. For example, to use PostgreSQL 14.3, the service becomes postgres:14.3.

The postgres image can accept some environment variables. For more details, see the documentation on Docker Hub.

Use PostgreSQL with the Shell executor

You can also use PostgreSQL on manually configured servers that are using GitLab Runner with the Shell executor.

First install the PostgreSQL server:

sudo apt-get install -y postgresql postgresql-client libpq-dev

The next step is to create a user, so sign in to PostgreSQL:

sudo -u postgres psql -d template1

Then create a user (in our case runner) which is used by your application. Change $password in the command below to a real strong password.

NOTE: Be sure to not enter template1=# in the following commands, as that's part of the PostgreSQL prompt.

template1=# CREATE USER runner WITH PASSWORD '$password' CREATEDB;

The created user has the privilege to create databases (CREATEDB). The following steps describe how to create a database explicitly for that user, but having that privilege can be useful if in your testing framework you have tools that drop and create databases.

Create the database and grant all privileges to it for the user runner:

template1=# CREATE DATABASE nice_marmot OWNER runner;

If all went well, you can now quit the database session:

template1=# \q

Now, try to connect to the newly created database with the user runner to check that everything is in place.

psql -U runner -h localhost -d nice_marmot -W

This command explicitly directs psql to connect to localhost to use the md5 authentication. If you omit this step, you are denied access.

Finally, configure your application to use the database, for example:

Host: localhost
User: runner
Password: $password
Database: nice_marmot

Example project

We have set up an Example PostgreSQL Project for your convenience that runs on GitLab.com using our publicly available instance runners.

Want to hack on it? Fork it, commit, and push your changes. Within a few moments the changes are picked by a public runner and the job begins.