aa-taskmonitor 0.23.1

An Alliance Auth app for monitoring celery tasks.

Task Monitor

An Alliance Auth app for monitoring celery tasks.

Contents

• Features
• Screenshots
• Installation
• Command line utility
• Settings
• FAQ
• Change Log

Features

Task Monitor enables administrators to monitor celery tasks running on their system.

• Creates a log of all recently executed celery tasks including failed and retried tasks.
• Stores many details in task logs to support the analysis of potential celery issues, including the parameters a task with called with and complete exception messages
• Keeps the storage needs in check by automatically deleting older task logs and removing likely bloat from the collected data (but can also be turned off)
• Admins can investigate task log with search & filters
• Admins can view details for each task log incl. exceptions and trace logs
• Admins can review reports with charts providing answers to common questions, e.g:
  • How many tasks have failed/retried?
  • How many tasks where run by each of my apps?
  • Which are the most frequent tasks?
  • Which tasks have the longest runtime?
  • Which tasks failed the most?
  • How much backlog do I have in task queue over time?
• Admins can export all task logs to a CSV file for further analysis with 3rd party tools (e.g. Google sheets)
• Admins can see detailed statistics for all tasks
• Command line utility to manage task logs and queue directly

Screenshots

Full log of all recently executed tasks

View details for each task incl. exception tracelogs

Example chart in reports

View detailed statistics for all tasks

Installation

Step 1 - Check prerequisites

Task Monitor is a plugin for Alliance Auth. If you don't have Alliance Auth running already, please install it first before proceeding. (see the official AA installation guide for details)

Step 2 - Install app

Make sure you are in the virtual environment (venv) of your Alliance Auth installation. Then install the newest release from PyPI:

pip install aa-taskmonitor

Step 3 - Configure Auth settings

Configure your Auth settings (local.py) as follows:

• Add 'taskmonitor' to INSTALLED_APPS
• Optional: Add additional settings if you want to change any defaults. See Settings for the full list.

Step 4 - Finalize App installation

Run migrations & copy static files

python manage.py migrate
python manage.py collectstatic --noinput

Restart your supervisor services for Auth.

Command line utility

You can manage your logs and queue also directly via a command line utility.

The basic syntax for using the utility is:

python manage.py taskmonitorctl {command} {target}

For example you can inspect your logs with:

python manage.py taskmonitorctl inspect logs

Or for example you can purge your task queue with:

python manage.py taskmonitorctl purge queue

To get an overview of available commands run:

python manage.py taskmonitorctl --help

To get an overview of available targets for a command run:

python manage.py taskmonitorctl inspect --help

Settings

Here is a list of available settings for this app. They can be configured by adding them to your AA settings file (local.py).

Note that all settings are optional and the app will use the documented default settings if they are not used.

Name	Description	Default
TASKMONITOR_APP_NAME_MAPPING_CONFIG	Ability to map tasks to the same app name. Map must be a dictionary with string keys and list of strings as value. All app names in the list will be replaced by it's key.	{}
TASKMONITOR_DATA_MAX_AGE	Max age of logged tasks in hours. Older logs be deleted automatically.	24
TASKMONITOR_DELETE_STALE_BATCH_SIZE	Size of task logs deleted together in one batch.	5000
TASKMONITOR_ENABLED	Global switch to enable/disable task monitor.	True
TASKMONITOR_HOUSEKEEPING_FREQUENCY	Frequency of house keeping runs in minutes.	30
TASKMONITOR_QUEUED_TASKS_ADMIN_LIMIT	The admin page will stop showing the list of queued tasks above this limit to protect against crashing caused by too high memory consumption.	100000
TASKMONITOR_QUEUED_TASKS_CACHE_TIMEOUT	Timeout for caching queued tasks in seconds. 0 will deactivate the cache.	10
TASKMONITOR_REPORTS_MAX_AGE	Max age of cached reports in minutes.	60
TASKMONITOR_REPORTS_MAX_TOP	Max items to show in the top reports. e.g. 10 will shop the top ten items.	20
TASKMONITOR_TRUNCATE_NESTED_DATA	Whether deeply nested task params and results are truncated.	True

FAQ

Is it possible to store task logs longer then for just 24 hours?

Yes, there is a setting, which you can increase according to your needs. However, please keep in mind that your storage needs will increase accordingly. The current approx. usage is 0.5 KB per entry, so e.g. you need approx. 500 MB to store 1.000.000 task logs. Note that you can use the command line utility to find out how much space your logs are currently using.

How is this app different from celery analytics?

Celery Analytics seams to be designed mainly as data source for reports on Grafana. It apparently works great if you want to integrate analysis about your task executions into a Grafana dashboards. But it's usability without Grafana is limited.

Task Monitor on the other hand aims to be fully functional standalone by providing reports and many useful features for analyzing your task logs directly on the admin site. It also provides a more complete picture, since Celery Analytics ignores retried tasks.

How is this app different from celery's flower?

Flower offers more detailed and technical information about task runs and might therefore be more most for developers. However, it not designed to store a larger number of task logs (default is only 10K) and is appears therefore to be less suited for Alliance Auth, where you typically have 100K+ tasks per day.

What does data truncating do exactly?

Task Monitor has data truncating enabled by default. It is applied when storing args, kwargs and results in task logs. This helps to reduce the storage consumptions and also makes the task log better readable. But it can be turned off.

Task args are truncated by clearing all nested containers.

Example: [1, [2, 3], 4] becomes [1, [], 4]

Task kwargs are truncated by clearing all nested containers in values.

Example: {"a": [1, 2], "b": 3} becomes {"a": [], "b": 3}

Finally, task results are truncated like args and kwargs depending on their type. In addition lists of empty containers are compressed.

Example: [ [], [], [] ] becomes []