tables_cleaner is a Django app used to remove oldest records from specific db tables.
It is intended to be used in production to keep under control the size of growing tables containing temporary data (used for logging, auditing, ...), but preserving the most recent records according to the constraints assigned by design (see TABLES_CLEANER_TABLES setting).
Installation:
pip install django-tables-cleaner or pip install git+https://github.com/morlandi/django-tables-cleaner
Add "tables_cleaner" to your INSTALLED_APPS setting like this:
INSTALLED_APPS = [ ... 'tables_cleaner', ]
Run the management command periodically (i.e. with cron)
python manage.py clean_tables
The first option is to recall periodically the management command clean_tables, for example via cron:
usage: manage.py clean_tables [-h] [--database DATABASE] [-d] [--vacuum] [--version] [-v {0,1,2,3}] [--settings SETTINGS] [--pythonpath PYTHONPATH] [--traceback] [--no-color] optional arguments: -h, --help show this help message and exit --database DATABASE Nominates a specific database to load fixtures into. Defaults to the "default" database. -d, --dry-run Don't actually delete records (default: False) --vacuum Run VACUUM after deletion --version show program's version number and exit -v {0,1,2,3}, --verbosity {0,1,2,3} Verbosity level; 0=minimal output, 1=normal output, 2=verbose output, 3=very verbose output --settings SETTINGS The Python path to a settings module, e.g. "myproject.settings.main". If this isn't provided, the DJANGO_SETTINGS_MODULE environment variable will be used. --pythonpath PYTHONPATH A directory to add to the Python path, e.g. "/home/djangoprojects/myproject". --traceback Raise on CommandError exceptions --no-color Don't colorize the command output.
Or, when using a different scheduling strategy (for example with django-cron) you can call from Python code the following function:
clean_tables(logger=None, dry_run=False)
For example:
import tables_cleaner
tables_cleaner.clean_tables()
Finally, for very specific needs, you can recall the real workhorse directly:
def clean_table(model_name, keep_records, keep_since_days, keep_since_hours, get_latest_by=None, logger=None, dry_run=False)
which act on a single table, and doesn't require any setting.
- TABLES_CLEANER_TABLES
The list of models to be cleaned;
options:
- keep_records: n. of most recent records to be preserved; 0=unused
- keep_since_days: always preserve records more recent than this; 0=unused
- keep_since_hourse: always preserve records more recent than this; 0=unused
Example:
TABLES_CLEANER_TABLES = [ { 'model_name': 'backend.log', 'keep_records': 1000, 'keep_since_days': 1, 'keep_since_hours': 0, }, { 'model_name': 'tasks.updatedevicetask', 'keep_records': 100, 'keep_since_days': 0, 'keep_since_hours': 12, 'get_latest_by': 'created', }, ]
get_latest_by attribute is optional; if not supplied, Model's Meta get_latest_by is used instead.
"VACUUM" is optionally executed as a final activity ('--vacuum').
Since version v0.1.0, we opted to use "VACUUM" instead of "VACUUM FULL", since that seems more appropriate for ordinary database maintenance, for the following reasons:
- it's available for Postgresql and Sqlite (and, hopefully, for other databases too)
- database owners are allowed to vacuum all tables in their databases
- an exclusive lock is not required
- it's potentially much faster
PostgreSQL documentation explicitly states that The FULL option is not recommended for routine use; see: VACUUM — garbage-collect and optionally analyze a database
Thanks to John Vandenberg for bringing my attention to this.
Removing rows in the database when the Model contains one or more FileField or ImageField is not enough, since some garbage is left in the Media folder.
I normally use django-cleanup to cope with this.
A few unit tests have been provided.
Prepare the virtual environment as follows:
python -m pip install -r requirements.txt
then:
./runtests.py
or:
coverage run --source='.' runtests.py coverage report
- Using the Django test runner to test reusable applications
- Advanced tutorial: How to write reusable apps
- TODO: Running tests using tox
This code is distributed under the terms of the MIT license.