Why?

I noticed that a relatively small number of sessions in my Django app were consuming 50+KiB of data each, and wanted to understand why. It isn’t feasible or appealing to analyze large numbers of Django sessions by hand, and I couldn’t find any existing tools that did batch analysis, so I wrote this tool.

Installation

Install the distribution from PyPI, e.g. pip install django-analyze-sessions

Then, add analyze_sessions to INSTALLED_APPS in your Django project.

That’s it. You can now run django-admin.py and confirm that the command shows up.

Example Uses

Break down the space usage of all sessions with more than 15KB of session data::

django-admin.py analyze-sessions --bigger-than=15000
Processed 2 sessions out of 5125 total sessions
Average size was 16885.50 bytes
Saw the following keys:
    foo (2 times, avg. size 15723.00 bytes)
    bar (2 times, avg. size 618.50 bytes)
    baz (2 times, avg. size 211.50 bytes)

Break down the space usage of all sessions whose keys are in the given input file (useful for when you want to select sessions on more than encoded size)::

django-admin.py analyze-sessions --ids-from=/tmp/ids-to-check
Processed 2 sessions out of 5125 total sessions
Average size was 16885.50 bytes
Saw the following keys:
    foo (2 times, avg. size 15723.00 bytes)
    bar (2 times, avg. size 618.50 bytes)
    baz (2 times, avg. size 211.50 bytes)

Known Issues and Bugs

• The custom WHERE clause used to filter on session size may only work on MySQL (I don’t have access to a Django installation on another DBMS, so I can’t vouch for how well it’ll work there).

• The tool could stand to be smarter about size abbreviations (e.g., by learning how to interpret 100KB, or abbreviating output)

• The packaging (such as it is) isn’t very good.

Author

Kevan Carstensen <kevan@isnotajoke.com>