Django PostgreSQL Copy

The package django-pg-copy provides Django management commands for backing up and restoring PostgreSQL databases. These were developed for copying production databases to development, to allow developers to share images with one another, or bring local development databases up to date. It can also be handy for creating different local databases for different branches, and for only creating one migration after tweaking models to get them the way they need to be.

We also use it with Jenkins to automatically back up production, and restore to a staging database environment, so we can test new migrations repeatedly to ensure they'll work when we run them in production.

Installation

pip install django-pg-copy

Then add 'pg_copy', to your INSTALLED_APPS list. It is recommended that it is used in all environments (development, production) so that you can use it against different instances of your database.

Settings

PG_COPY_BACKUP_PATH = 'db_backup'

By default, PostgreSQL backups will be stored in a directory called db_backup where you run the command. This setting will override that location.

It is also recommended to add this path to your .gitignore file, if the path falls under your version control repository.

Parameters

• --database: The database defined in the DATABASES settings to backup from or restore to.
• --db-override: A value to override the db argument sent to psql.
• --host-override: A value to override the host argument sent to psql.
• --pg-home: The path to the PostgreSQL installation, if it is not on your path.
• --file, -f: The filename to backup to, or restore from.
• --jobs, -j: The number of parallel jobs to run. This can drastically increase the speed of the backup and restore. When backing up, this must use the --directory option. Be careful! This will also create multiple database connections and can slow your database down.
• --directory, -d: Restore from a directory. Overrides --file when used.

Backup Only Parameters for pg_backup

• --ignore-table, -i: Excludes the table completely during the backup process. Can pass multiple tables: -i bigtable1 -i bigtable2
• --exclude-table-data, -e: Excludes the table data during the backup process. Can pass multiple tables: -e bigtable1 -e bigtable2

Restore Only Parameters for pg_restore

• --no-confirm: Restores the database without confirmation: be careful! DON'T OVERWRITE YOUR PROD DATABASE! (Ask me how I know.)

Example Commands

python manage.py pg_backup

This command will create a backup file in the directory ./db_backup/ (or the directory you specified with PG_COPY_BACKUP_PATH) called [timestamp].sqlc using the default settings from DATABASES using the default Django settings file resolved by manage.py.

python manage.py pg_backup --settings=config.settings.production --database=default --filename=my_backup.sqlc

This command will create a backup in the same directory as manage.py called my_backup.sqlc using the default settings from DATABASES using the Django settings file located at config/settings/production.py.

python manage.py pg_backup --directory=/tmp/pg_backup --jobs=8

This command will create a backup directory at /tmp/backup use 8 parallel jobs and database connections.

python manage.py pg_backup -i bigtable1 -i bigtable2

This will do the same as the previous command, but omit the tables named bigtable1 and bigtable2.

python manage.py pg_backup -e bigtable1 -e bigtable2

This will do the same as the previous command, but include the table structure without any data for the tables named bigtable1 and bigtable2.

python manage.py pg_restore

This command will provide a list of backup files in PG_COPY_BACKUP_PATH that can be restored. After selecting a backup file, it will confirm that the user wants to overwrite the destination database by showing which server and database will be overwritten from the settings. Here's what it will look like:

(venv) [django-project]$ ./manage.py pg_restore
There are 8 backup files in '/var/dev/username/django-project/db_backup'. Which would you like to restore?
1: 2020-05-27-13-33-38.sqlc
2: 2020-05-10-09-07-45.sqlc
3: 2020-05-11-13-05-49.sqlc
4: 2020-07-14-15-21-22.sqlc
5: 2020-05-15-08-31-59.sqlc
6: 2020-06-11-13-11-00.sqlc
7: 2020-06-02-13-28-09.sqlc
8: 2020-05-27-14-54-15.sqlc
Enter number of the file to restore: 4
About to restore 'django_project_db' on host 'localhost' from the file:
'/var/dev/username/django-project/db_backup/2020-07-14-15-21-22.sqlc'.
THIS WILL OVERWRITE THE DATABASE.
Type "yes" to start the restore [no]: yes

python manage.py pg_restore --filename=my_file.sqlc --no-confirm

This command will read the file my_file.sqlc and skip confirmation that the user wants to overwrite the destination database.

python manage.py pg_restore --directory=/tmp/pg_backup --jobs=8

This command will restore from the directory /tmp/pg_backup using 8 parallel jobs and database connections.

Known Issues

Errors Displayed During pg_restore

When restoring, PostgreSQL's pg_restore command may output some warnings and erorrs. I haven't figured out a command line option to make these warnings disappear, but they can likely be ignored if you read them. Here is an example of what these errors may look like:

pg_restore: [archiver (db)] Error while PROCESSING TOC:
pg_restore: [archiver (db)] Error from TOC entry 1; 3079 13792 EXTENSION plpgsql
pg_restore: [archiver (db)] could not execute query: ERROR:  must be owner of extension plpgsql
    Command was: DROP EXTENSION IF EXISTS plpgsql;

pg_restore: [archiver (db)] Error from TOC entry 6; 2615 2200 SCHEMA public postgres
pg_restore: [archiver (db)] could not execute query: ERROR:  must be owner of schema public
    Command was: DROP SCHEMA IF EXISTS public;

pg_restore: [archiver (db)] could not execute query: ERROR:  schema "public" already exists
    Command was: CREATE SCHEMA public;

These may be errors from extensions that aren't used, or warnings about default schemata that already exist.

Release Notes

Release notes are available on GitHub.

Maintainer

• Timothy Allen at The Wharton School

This package is maintained by the staff of Wharton Research Data Services. We are thrilled that The Wharton School allows us a certain amount of time to contribute to open-source projects. We add features as they are necessary for our projects, and try to keep up with Issues and Pull Requests as best we can. Due to constraints of time (our full time jobs!), Feature Requests without a Pull Request may not be implemented, but we are always open to new ideas and grateful for contributions and our package users.

Contributors

• Alex Malek (https://github.com/amalek215)
• Noel Victor (https://github.com/noeldvictor)
• Ryan Sullivan (https://github.com/rgs258)