Ansible role: MariaDB

Install and configure MariaDB Server on Debian/Ubuntu.

Optionally, this role also permits one to:

• deploy a primary/replica cluster;
• setup backups and rotation of the dumps.

Requirements

The role uses community.mysql.mysql_user and community.mysql.mysql_db ansible modules that depend on PyMySQL. Only Python 3.X versions are supported and tested. Python 2.7 is probably working but not tested.

If you need to deploy a cluster then you need Ansible 2.12+ since there was a change in the community.mysql.mysql_replication naming for cluster node (primary/replica is now preferred).

pre-commit

Any code submitted to this project is checked with the pre-commit framework. To make sure that your code will pass the checks, you can execute the pre-commit checks locally before "git pushing" your code.

Here is how:

❯ make venv
❯ source .venv/bin/activate
.venv ❯ make install-pre-commit
.venv ❯ make pre-commit-run

You can also install the pre-commit tool so that any commit will be checked automatically.

Testing (Molecule)

The role is tested with the Molecule project. By default this role will be tested with the podman driver, but you can easily adapt it to use docker if you prefer.

Here is an example how you can test a deployment of MariaDB Server 10.6 packaged by the MariaDB Foundation (MDBF) on Almalinux 9:

❯ make install
❯ source .venv/bin/activate
.venv ❯ export MOLECULE_DISTRO=almalinux-9
.venv ❯ export MOLECULE_PLAYBOOK=converge-mdbf.yml
.venv ❯ export MARIADB_VERSION="10.6"
.venv ❯ molecule test
...

And here is another example how you can test a deployment of a MariaDB Server cluster (3 nodes: 1 primary, 2 replica) on fedora 37:

.venv ❯ export MOLECULE_DISTRO=fedora-37
.venv ❯ molecule test -s cluster
...

Fact gathering (performance)

Fact gathering is only needed if you want to use the MariaDB official repository (https://mariadb.org/download/?t=repo-config)

Role variables

Available variables are listed below, along with default values (see defaults/main.yml).

MariaDB version

mariadb_use_official_repo: false
mariadb_use_official_repo_url: https://deb.mariadb.org
mariadb_use_official_repo_version: "10.10"

You may deploy the MariaDB Server version that comes with your distribution (Debian/Ubuntu) or deploy the version packaged by the MariaDB Foundation. You can use the MariaDB Foundation repository configuration tool: https://mariadb.org/download/#mariadb-repositories

By default, we deploy the MariaDB Server version that comes with the distribution.

MariaDB service activation and restart

mariadb_enabled_on_startup: true
mariadb_can_restart: true

Warning: you may consider setting mariadb_can_restart to false on production systems to prevent ansible runs from restarting the MariaDB Server.

General configuration

To populate the MariaDB Server configuration file, we use almost only raw variables. This permits more flexibility and a very simple templates/mariadb.cnf.j2 file.

By default, some common and standard options are deployed based on the MariaDB Foundation package. Those default values are only meant as an example and for testing deployments and you are encouraged to use your own values.

Basic settings

default value depends on OS means that the value is overridden at OS level, see vars.

mariadb_config_file: "default value depends on OS"
mariadb_data_dir: "default value depends on OS"
mariadb_port: 3306
mariadb_bind_address: 127.0.0.1
mariadb_unix_socket: "default value depends on OS"

mariadb_basic_settings_raw: |
  user                  = mysql
  pid-file              = {{ mariadb_pid_file }}
  socket                = {{ mariadb_unix_socket }}
  basedir               = /usr
  datadir               = {{ mariadb_data_dir }}
  tmpdir                = /tmp
  lc-messages-dir       = /usr/share/mysql
  lc_messages           = en_US
  skip-external-locking
  port                  = {{ mariadb_port }}
  bind-address          = {{ mariadb_bind_address }}

Fine-tuning

mariadb_fine_tuning_raw: |
  max_connections         = 100
  connect_timeout         = 5
  wait_timeout            = 600
  max_allowed_packet      = 16M
  thread_cache_size       = 128
  sort_buffer_size        = 4M
  bulk_insert_buffer_size = 16M
  tmp_table_size          = 32M
  max_heap_table_size     = 32M

Query cache

mariadb_query_cache_raw: |
  query_cache_size        = 16M

Logging

mariadb_logging_raw: |
  log_error = "default value depends on OS"

Character sets

mariadb_character_sets_raw: |
  character-set-server = utf8mb4
  collation-server     = utf8mb4_general_ci

InnoDB

mariadb_innodb_raw: |
  # InnoDB is enabled by default with a 10 MB datafile in /var/lib/mysql/.
  # Read the manual for more InnoDB related options. There are many!

Mariadbdump

mariadb_mysqldump_raw: |
  quick
  quote-names
  max_allowed_packet = 16M

Database management

mariadb_databases: []
#   - name: db1
#     collation: utf8_general_ci
#     encoding: utf8
#     replicate: true|false

See: https://docs.ansible.com/ansible/latest/modules/mysql_db_module.html

User management

mariadb_users: []
#   - name: user
#     host: 100.64.200.10
#     password: password
#     priv: "*.*:USAGE/db1.*:ALL"
#     state: present|absent

See: https://docs.ansible.com/ansible/latest/modules/mysql_user_module.html

Replication (optional)

Replication is only enabled if mariadb_replication_role has a value (primary or replica).

The replication setup on the replica use the GTID autopositioning, see https://mariadb.com/kb/en/library/change-master-to/#master_use_gtid

Common vars

# Same keys as `mariadb_users` above.
# priv is set to "*.*:REPLICATION SLAVE" by default
mariadb_replication_user: []

Primary node variables

mariadb_replication_role: primary
mariadb_server_id: 1
mariadb_max_binlog_size: 100M
mariadb_binlog_format: MIXED
mariadb_expire_logs_days: 10

Replica variables

mariadb_replication_role: replica
mariadb_server_id: 1
mariadb_replication_primary_ip: IP

Backups (optional)

# db dumps backup
mariadb_backup_db: false
mariadb_backup_db_cron_min: 50
mariadb_backup_db_cron_hour: 00
mariadb_backup_db_dir: /mnt/backup
mariadb_backup_db_rotation: 15
# set to "1>" to get only STDERR on cron
mariadb_backup_cron_std_output: "2>&1 | tee"

# name of the database to dump
# (mandatory if mariadb_backup_db is set to true)
mariadb_backup_db_name: []
#   - db1
#   - db2

Database dumps are done serially and the compression step (gzip) is done after to avoid lengthy locks.

Example playbook

- hosts: db
  roles:
    - fauust.mariadb

Lincense

GNU General Public License v3.0