This image selective-dumps your postgres databases every hour (OR custom cron defined in BACKUP_CRON_SCHEDULE), compresses the dump using zip and uploads it to an amazon S3 bucket. Backups older than 30 days (OR days defined in AWS_KEEP_FOR_DAYS) are deleted automatically.

It creates a group of .csv files, a schema.sql for database schema definition and a script to import all.

To import, use: PGPASSWORD=password PGUSER=user ./import_script.sh

Configure the backup source and s3 target with these environment variables:

• AWS_REGION (for example eu-central-1)
• AWS_ACCESS_KEY_ID
• AWS_SECRET_ACCESS_KEY
• AWS_BUCKET_NAME
• AWS_KEEP_FOR_DAYS
• BACKUP_PATH
• BACKUP_CRON_SCHEDULE
• MASTER_OBJ_TBL main table to export
• MASTER_OBJ_IDS objects' ids of main table to export
• MASTER_OBJ_CLMN main table reference column in (see below)
• MASTER_CHILD_TABLES childrens tables referenced by main table to export
• STANDARD_TABLES other tables to export
• PGHOST
• PGDATABASE
• PGPORT
• PGUSER
• PGPASSWORD

Example: This is our schema in pg database: users(id, name, age) blogs(id, name, description) comments(id, user_id, blog_id, text, timestamp) images(id, blog_id, url)

We want to export blogs, users and users' related comments. We don't want images. But we want a SELECTIVE dump, because we have 10k users and 1mln of comments in total, so we want to limit to users' ids 1 and 2 To export, assuming database name is test and admin user is postgres, we lauch: PGDATABASE=test PGUSER=postgres MASTER_OBJ_TBL="users" MASTER_OBJ_IDS="1, 2" MASTER_OBJ_CLMN="user_id" MASTER_CHILD_TABLES="comments" STANDARD_TABLES="blogs" ./export_script.sh

Now, a schema.sql, .csv files will be created and a file called import_script.sh. All of this will be zipped into backup.zip adn uploaded to S3 storage.