Setting up Databases and Services

This guide covers setting up the most popular databases and other services in the Travis CI environment.

You can check databases and services availability in the build environment you are using here.

All services use default settings, with the exception of some added users and relaxed security settings.

Starting Services #

Travis CI environments do not start services by default, to make more RAM available to build scripts. Start services by adding them to the services: section of your .travis.yml:

services: mongodb

If you install a service in the addons: section, such as MariaDB, you do not need to add it to the services: section as well.

To start several services:

  - riak
  - rabbitmq
  - memcached

If you download and install a service manually, you also have to start it in a before_install step. The services key only works for services we provision.


Start MySQL in your .travis.yml:

  - mysql

MySQL binds to and a socket defined in ~travis/.my.cnf and requires authentication. You can connect using the username travis or root and a blank password.

Note that the travis user does not have the heightened privileges that the root user does.

  Ubuntu Precise Ubuntu Trusty Ubuntu Xenial Ubuntu Bionic Ubuntu Focal Ubuntu Jammy
MySQL 5.5.x 5.6.x 5.7.x 5.7.x 8.0.x 8.0.x

You can also install MySQL 5.7 on Ubuntu Trusty.

Using MySQL with ActiveRecord #

config/database.yml example for Ruby projects using ActiveRecord:

  adapter: mysql2
  database: myapp_test
  username: travis
  encoding: utf8

You might have to create the myapp_test database first, for example in the before_install step in .travis.yml:

  - mysql -e 'CREATE DATABASE myapp_test;'

Note on test database #

In older versions of MySQL, the Ubuntu package provided the test database by default. This is no longer the case as of version 5.5.37 due to security concerns (See change log).

The test database may be created if needed, for example in the before_install step in .travis.yml:

  - mysql -e 'CREATE DATABASE IF NOT EXISTS test;'

MySQL 5.7 #

MySQL 5.7 is the default on the Xenial (dist: xenial) and Bionic (dist: bionic) images.

Since July 21st 2019, MySQL 5.7 is not supported on Ubuntu Trusty (14.04) anymore. See MySQL Product Support EOL Announcements and this post in the MySQL Forums.

PostgreSQL #

Start PostgreSQL in your .travis.yml:

  - postgresql

Using PostgreSQL in your Builds #

The default user for accessing the local PostgreSQL server is postgres with a blank password.

Create a database for your application by adding a line to your .travis.yml:

  - psql -c 'create database travis_ci_test;' -U postgres

For a Rails application, you can now use the following database.yml configuration to access the database locally:

  adapter: postgresql
  database: travis_ci_test

If your local test setup uses different credentials or settings to access the local test database, we recommend putting these settings in a database.yml.travis in your repository and copying that over as part of your build:

  - cp config/database.yml.travis config/database.yml

Using a different PostgreSQL Version #

The Travis CI build environments use version 9.2 by default on Trusty images, but other versions from the official PostgreSQL APT repository are also available. To use a version other than the default, specify only the major.minor version in your .travis.yml:

  postgresql: "9.4"

Many PostgreSQL versions have been preinstalled in our build environments, and others may be added and activated at build time by using a combination of the postgresql and apt addons along with a global env var override for PGPORT and for PGUSER:

  postgresql: "11"
    - postgresql-11
    - postgresql-client-11
  - PGPORT=5433
  - PGUSER=travis

In the Xenial images Postgres 9.4 through 9.6 just need the version specified and use the user postgres by default and the default port of 5432.

For PostgreSQL 10 you must specify the packages to install it and the user is postgres and the port is 5432. For PostgreSQL 11 and 12 you must specify the packages, but the user is travis and the port is 5433 instead. So you must specify the PGPORT

Using PostGIS #

Install the version of PostGIS that matches your PostgreSQL version, and activate the PostGIS extension using:

  postgresql: 9.6
    - postgresql-9.6-postgis-2.3
  - psql -U postgres -c "create extension postgis"

PostgreSQL and Locales #

The Travis CI build environment comes with a number of pre-installed locales, but you can also install additional ones, should you require them.

Installing Locales #

The following example shows the lines you need to add to your .travis.yml to install the Spanish language pack.

Note that you need to remove the PostgreSQL version from the addons section of your .travis.yml:

  - sudo apt-get update
  - sudo apt-get install language-pack-es
  - sudo /etc/init.d/postgresql stop
  - sudo /etc/init.d/postgresql start 9.3

Using pg_config #

If your builds rely on the pg_config command, you need to install an additional apt package postgresql-server-dev-X.Y, where X.Y matches the version of PostgreSQL you are using.

For example:

  postgresql: '9.4'
      - postgresql-server-dev-9.4

See this GitHub issue for additional details.

MariaDB #

MariaDB is a community-developed fork of MySQL. It is available as an addon on Travis CI.

To use MariaDB, specify the “major.minor” version you want to use in your .travis.yml. Versions are listed on the MariaDB web page.

  mariadb: '10.0'

The version number is exported as the TRAVIS_MARIADB_VERSION environment variable.

SQLite3 #

The easiest and simplest relational database.

SQLite3 in Ruby Projects #

Add the sqlite3 ruby bindings to your bundle:

# Gemfile
# for CRuby, Rubinius, including Windows and RubyInstaller
gem "sqlite3", :platform => [:ruby, :mswin, :mingw]

# for JRuby
gem "jdbc-sqlite3", :platform => :jruby

If you use ActiveRecord, add the following to your config/database.yml:

  adapter: sqlite3
  database: ":memory:"
  timeout: 500

Or if you’re not using a config/database.yml, connect to the database manually:

ActiveRecord::Base.establish_connection :adapter => 'sqlite3',
                                        :database => ':memory:'

MongoDB #

Start MongoDB in your .travis.yml:

  - mongodb

MongoDB binds to and requires no authentication or database creation up front. If you add an admin user authentication is enabled, since mongod is started with the --auth argument.

Note: Admin users are users created in the admin database.

To create users for your database, add a before_script section to your .travis.yml:

  - mongo mydb_test --eval 'db.createUser({user:"travis",pwd:"test",roles:["readWrite"]});'

MongoDB does not immediately accept connections #

A few users have reported that MongoDB does not accept connections when from the build script.

The issue is intermittent, and the only reliable way to avoid it is to inject an artificial wait before making the first connection:

Add the following before_script to your .travis.yml to wait before connecting to MongoDB:

  - sleep 15
  - mongo mydb_test --eval 'db.createUser({user:"travis",pwd:"test",roles:["readWrite"]});'

CouchDB #

Start CouchDB in your .travis.yml:

  - couchdb

CouchDB binds to, uses default configuration on dist:xenial and earlier Linux distributions and does not require authentication (in CouchDB terms it runs in admin party).

However for bionic, authentication is required with username admin and password travis e.g. curl -X PUT http://admin:travis@localhost:5984/<db_name>.

Before using CouchDB you need to create the database as part of your build process:

  - curl -X PUT localhost:5984/myapp_test

RabbitMQ #

RabbitMQ requires setuid flags, so you can only run RabbitMQ as a service on macOS or Ubuntu Trusty infrastructure.

Start RabbitMQ in your .travis.yml:

  - rabbitmq

RabbitMQ uses the default configuration:

  • vhost: /
  • username: guest
  • password: guest

You can set up more vhosts and roles in the before_script section of your .travis.yml.

RabbitMQ can be launched on Ubuntu Xenial using the APT addon in .travis.yml:

    - rabbitmq-server 

Riak #

Riak is only available in the Ubuntu Trusty environment.

Start Riak in your .travis.yml:

  - riak

Riak uses the default configuration with Bitcask as storage backend.

Riak Search is deactivated by default.

Memcached #

Start Memcached service in your .travis.yml:

  - memcached

Memcached uses the default configuration and binds to localhost.

Redis #

Start Redis in your .travis.yml:

  - redis-server

Redis uses the default configuration and is available on localhost.

Cassandra #

Start Cassandra in your .travis.yml:

  - cassandra

Cassandra is downloaded from the Apache apt repository and uses the default configuration. It is available on

Installing older versions of Cassandra #

Use the following example to install a specific older version of Cassandra in your .travis.yml:

  - sudo rm -rf /var/lib/cassandra/*
  - wget && tar -xvzf apache-cassandra-1.2.18-bin.tar.gz && sudo sh apache-cassandra-1.2.18/bin/cassandra

Neo4j #

Start Neo4j in your .travis.yml:

  - neo4j

Neo4j Server uses default configuration and binds to localhost on port 7474.

ElasticSearch #

Start ElasticSearch in your .travis.yml:

  - elasticsearch

ElasticSearch takes few seconds to start, to make sure it is available when the build script runs add a small delay to your build script:

  - sleep 10

ElasticSearch uses the default configuration and is available on

Installing specific versions of ElasticSearch #

You can overwrite the installed ElasticSearch with the version you need (e.g., 7.6.2) with the following:

  - curl -o elasticsearch.deb
  - sudo dpkg -i --force-confnew elasticsearch.deb
  - sudo chown -R elasticsearch:elasticsearch /etc/default/elasticsearch
  - sudo service elasticsearch restart

We advise verifying the validity of the download URL on ElasticSearch’s website.

Truncated Output in the Build Log #

When ElasticSearch starts, you may see a truncated error message such as:

$ sudo service elasticsearch start
 * Starting ElasticSearch Server       ission denied on key 'vm.max_map_count'

This is due to a recent change in ElasticSearch, as reported here. The message is harmless, and the service is functional.

RethinkDB #

To use RethinkDB with Travis CI, list it as an addon in the .travis.yml configuration file, specifying the version number as a string.

  rethinkdb: '2.3.4'

If you specify a partial version number, the addon will install and run the latest version that matches. For example, '2.3' will match the latest RethinkDB version in the 2.3.x line.

Two environment variables are exported:

  • TRAVIS_RETHINKDB_VERSION is the version specified in the configuration (e.g., '2.3.4', or '2.3').
  • TRAVIS_RETHINKDB_PACKAGE_VERSION is the full version of the package that was installed (e.g., '2.3.4+1~0precise').

When enabled, RethinkDB will start on localhost at the default port (28015).

Multiple Database Builds #

If you need to run multiple builds using different databases, you can configure environment variables and a before_script or before_install line to create a build matrix.

Using environment variables and a before_script step #

Use the DB environment variable to specify the name of the database configuration. Locally you would run:

DB=postgres [commands to run your tests]

On Travis CI you want to create a build matrix of three builds each having the DB variable exported with a different value, and for that you can use the env option in .travis.yml:

  - DB=sqlite
  - DB=mysql
  - DB=postgres

Then you can use those values in a before_install (or before_script) step to set up each database. For example:

  - sh -c "if [ '$DB' = 'postgres' ]; then psql -c 'DROP DATABASE IF EXISTS tests;' -U postgres; fi"
  - sh -c "if [ '$DB' = 'postgres' ]; then psql -c 'DROP DATABASE IF EXISTS tests_tmp;' -U postgres; fi"
  - sh -c "if [ '$DB' = 'postgres' ]; then psql -c 'CREATE DATABASE tests;' -U postgres; fi"
  - sh -c "if [ '$DB' = 'postgres' ]; then psql -c 'CREATE DATABASE tests_tmp;' -U postgres; fi"
  - sh -c "if [ '$DB' = 'mysql' ]; then mysql -e 'CREATE DATABASE IF NOT EXISTS tests_tmp; CREATE DATABASE IF NOT EXISTS tests;'; fi"

Travis CI does not have any special support for these variables, it just creates three builds with different exported values. It is up to your build script and before_install or before_script steps to make use of them.

For a real example, see doctrine/doctrine2 .travis.yml.

Using Ruby #

Another approach is put all database configuration in one YAML file (test/database.yml for example), like ActiveRecord does:

  adapter: sqlite3
  database: ":memory:"
  timeout: 500
  adapter: mysql2
  database: myapp_test
  encoding: utf8
  adapter: postgresql
  database: myapp_test
  username: postgres

Then, in your test suite, read that data into a configurations hash:

configs = YAML.load_file('test/database.yml')
ActiveRecord::Base.configurations = configs

db_name = ENV['DB'] || 'sqlite'
ActiveRecord::Base.default_timezone = :utc