services/roles/restic-s3
2022-01-02 12:54:29 +01:00
..
defaults feat(restic-s3): add ansible role for managing backups using restic 2022-01-02 12:54:29 +01:00
handlers feat(restic-s3): add ansible role for managing backups using restic 2022-01-02 12:54:29 +01:00
tasks feat(restic-s3): add ansible role for managing backups using restic 2022-01-02 12:54:29 +01:00
templates feat(restic-s3): add ansible role for managing backups using restic 2022-01-02 12:54:29 +01:00
README.md feat(restic-s3): add ansible role for managing backups using restic 2022-01-02 12:54:29 +01:00

finallycoffee.services.restic-s3

Ansible role for backup up data using restic to an s3-compatible backend, utilizing systemd timers for scheduling

Overview

The s3 repository and the credentials for it are specified in restic_repo_url, restic_s3_key_id and restic_s3_access_key. As restic encrypts the data before storing it, the restic_repo_password needs to be populated with a strong key, and saved accordingly as only this key can be used to decrypt the data for a restore!

Backing up data

A job name like $service-postgres or similar needs to be set in restic_job_name, which is used for naming the systemd units, their syslog identifiers etc.

If backing up filesystem locations, the paths need to be specified in restic_backup_paths as lists of strings representing absolute filesystem locations.

If backing up f.ex. database or other data which is generating backups using a command like pg_dump, use restic_backup_stdin_command (which needs to output to stdout) in conjunction with restic_backup_stdin_command_filename to name the resulting output (required).

Policy

The backup policy can be adjusted by overriding the restic_policy_keep_* variables, with the defaults being:

restic_policy_keep_all_within: 1d
restic_policy_keep_hourly: 6
restic_policy_keep_daily: 2
restic_policy_keep_weekly: 7
restic_policy_keep_monthly: 4
restic_policy_backup_frequency: hourly

Note: restic_policy_backup_frequency must conform to systemds OnCalendar syntax, which can be checked using systemd-analyze calender $x.

Role behaviour

Per default, when the systemd unit for a job changes, the job is not immediately started. This can be overridden using restic_start_job_on_unit_change: true, which will immediately start the backup job if it's configuration changed.

The systemd unit runs with restic_user, which is root by default, guaranteeing that filesystem paths are always readable. The restic_user can be overridden, but care needs to be taken to ensure the user has permission to read all the provided filesystem paths / the backup command may be executed by the user.

If ansible should create the user, set restic_create_user to true, which will attempt to create the restic_user as a system user.

Installing

For Debian and RedHat, the role attempts to install restic using the default package manager's ansible module (apt/dnf). For other distributions, the generic package module tries to install restic_package_name (default: restic), which can be overridden if needed.