PostgreSQL S3 Backup

This Docker container enables you to backup a PostgreSQL database to AWS S3. The container dumps the database using pg_dump, creates a gzipped tarball of the dump, and uploads the tarball to S3.

The container does not handle the scheduling of the backup. The actual backup has to be invoked from an external service like cron, a cron Docker container, Jenkins, etc.

Usage

Credentials and connection details are required to be passed into the container through environment variables. This enables you to create your derivant of the image for your specific use case. The following environment variables are expected:

• PG_HOST (optional, default db)
• PG_PORT (optional, default 5432)
• PG_USER (optional, default postgres)
• PG_PASS (optional, no default value)
• AWS_REGION
• AWS_ACCESS_KEY
• AWS_SECRET_ACCESS_KEY

To run the backup for a locally deployed database, link the database container into the backup container:

$> docker run --link db_container:db \
              -e AWS_REGION=eu-central-1 \
              -e AWS_ACCESS_KEY_ID=12345 \
              -e AWS_SECRET_ACCESS_KEY=qwertz \
              lawitschka/postgresql-s3-backup \
              s3_bucket/path/in/bucket example_db

This will dump the database example_db into an archive called example_db-XXXXXXXXXX.sql.tar.gz. The XXXXXXXXXX will be replaced with a UNIX epoch at the time of the backup. The backup script then uploads the archive to a S3 bucket called s3_bucket and into the folder path/in/bucket.

Alternatively, the database host can be defined through usage of the PG_ environment variables:

$> docker run -e AWS_REGION=eu-central-1 \
              -e AWS_ACCESS_KEY_ID=12345 \
              -e AWS_SECRET_ACCESS_KEY=qwertz \
              -e PG_HOST=db.example.com \
              -e PG_USER=nondefaultuser \
              lawitschka/postgresql-s3-backup \
              s3_bucket/path/in/bucket example_db