Distributed error reporting

[akolson]

Background

Kolibri is primarily used in disconnected or low connectivity environments. In contrast to many offline platforms where error reporting can happen via Sentry or similar services, in these offline environments, errors that happen to users cannot be directly reported.

Kolibri currently offers the ability to copy and paste error reports in the frontend to share them back via other means - it would be useful to automate this process and allow error reports to be automatically shared back when connectivity is available.

Story

As a user of Kolibri in a low connectivity environment, I want any errors I encounter while using the platform to be automatically queued and sent once connectivity is restored, so that they are addressed without requiring manual intervention.

Requirements

• Django error reporting mechanisms should be used to capture and report backend errors.
• An appropriate error handling mechanism should be implemented on the frontend to report errors.
• An API to capture and report frontend errors should be implemented.
• A consumable error reporting stack should be used so that errors are only synced back once (See pingback mechanism).
• A database, external to Kolibri, should be used to store all reported errors. Using an external database ensures comprehensive error reporting beyond the confines of the codebase.
• It is worth looking into Kolibri's Sentry plugin for inspiration on how it captures errors on the frontend and backend*

Out of Scope

• Implementing changes to enable reporting of captured errors via the LE Telemetry API falls outside the scope of this task, as the LE Telemetry project is closed source.

Contributors

• Technical lead: @akolson
• Individual contributors: @thesujai
• Other relevant LE team members: @rtibbles

Feature Tasks

Beta Give feedback

1. Setup the database that stores all the errors raised during runtime #12244
   DEV: backend gsoc +2
   
2. Implement the ErrorReport model #12257
   DEV: backend gsoc +2
   
3. Create a middleware to catch backend exceptions #12258
   DEV: backend gsoc +2
   
4. Implement the /api/errorreports/report endpoint #12259
   DEV: backend gsoc +2
   
5. Add the runtime error tracking mechanism for frontend #12284
   DEV: backend gsoc +2
   
6. Add a task in errorreports that reports all the errors to our telemetry server #12356
   DEV: backend gsoc +2
   
7. Update the ErrorReports model to have more context on errors #12376
   DEV: backend gsoc +2
   
8. Implement /api/v1/errors/report/ in telemetry #12472
   DEV: backend gsoc +2
9. Turn distributed error reporting into a plugin and implement error capture into Kolibri's Tasks #12473
   DEV: backend gsoc +2
   
10. Prevent sensitive information like user passwords from being logged/captured #12555
    DEV: backend gsoc +2
11. Move avg_request_time_to_error to the backend context #12556
    DEV: backend gsoc +2
    

[akolson]

New feature to be implemented in Telemetry

• https://github.com/learningequality/nutritionfacts/issues/39