[Feature Request] Add Native Azure Blob Storage Support

[Tewatia5355]

Summary

I'd like to propose adding native Azure Blob Storage support to Tandoor to enable proper image storage for Azure deployments (Azure Container Apps, Azure Web Apps).

Problem Statement

Currently, Tandoor only supports S3-compatible storage backends. However:

• Azure Blob Storage is NOT S3-compatible (despite what some sources suggest)
• Users deploying on Azure Container Apps experience image upload issues:
  • Images appear to upload successfully but return null URLs
  • Images are stored in ephemeral container storage (lost on restart)
  • No persistence across container deployments
• The official Docker image lacks azure-storage-blob dependencies
• No existing issues or feature requests address this (I've searched thoroughly)

Proposed Solution

Add native Azure Blob Storage support using django-storages Azure backend, following the same patterns as the existing S3 implementation.

Technical Approach

1. Dependencies: Add django-storages[azure]>=1.14.6 to requirements.txt
2. Configuration: Use Django's modern STORAGES configuration (Django 5.0+)
3. Environment Variables: Follow S3 naming pattern but with Azure-specific settings
4. Authentication: Support both account key and connection string methods
5. Backward Compatibility: No impact on existing S3 users

Example Configuration

# Required
AZURE_ACCOUNT_NAME=mystorageaccount
AZURE_ACCOUNT_KEY=base64encodedkey==
AZURE_CONTAINER=media

# Optional
AZURE_SSL=True
AZURE_URL_EXPIRATION_SECS=3600
AZURE_CUSTOM_DOMAIN=cdn.example.com

Community Benefits

• First Azure support for Tandoor (no existing solutions found)
• Enables Azure cloud deployments (Container Apps, Web Apps)
• Large user base impact - Azure is a major cloud provider
• No breaking changes to existing functionality
• Modern cloud-native approach

Implementation Plan

I've prepared a comprehensive technical implementation plan covering:

• Precise Django settings configuration
• Environment variable specifications
• Testing strategy (pytest-django + real Azure integration)
• Documentation updates
• Migration guidance

Estimated Timeline: 2 weeks (development + testing + documentation)

My Background

• Experienced with Django and Azure deployments
• Currently experiencing this exact issue in production
• Committed to following Tandoor's contributor guidelines
• Will sign CLA and follow all code standards (flake8, yapf, isort)

Questions for the Community

• Is this approach acceptable for Tandoor's architecture?
• Any preferred patterns for cloud storage integration?
• Should this include static files or only media files?
• Any concerns about additional dependencies?
• Would maintainers support this feature addition?

I'm happy to adjust the approach based on community feedback and would love to contribute this functionality to help other Azure users!

References

Django-storages Azure Backend
Tandoor Configuration Docs
Tandoor Contributor Guidelines

[vabene1111]

hi, thanks for the well written technical draft. For me this sounds good and I can imagine merging this into tandoor.

Please try to keep the complexity as low as possible.

The thing I remember was the most difficult with S3 was having the urls for the view in such a way that they allow caching, if you need any input here about the S3 implementation let me know.

Also I think that doing only mediafiles is the correct approach, as staticfiles dont take up much space and can then be handled by the CDN/webserver easily.

[Tewatia5355]

Thank you so much for the positive feedback and approval! I'm excited to work on this.

I completely agree with your guidance:

• Keep complexity low - I'll follow the existing S3 patterns as closely as possible to maintain consistency
• Media files only - Perfect, that simplifies the implementation and aligns with best practices
• URL caching considerations - I'd really appreciate any insights you have about the S3 URL implementation, especially around caching. I want to ensure the Azure implementation handles this properly from the start.

I've done some research on the S3 caching implementation in the repository and found the QUERYSTRING_AUTH and QUERYSTRING_EXPIRE settings. I plan to map these directly to Azure's equivalent settings (expiration_secs and SAS token handling) to keep the behavior consistent.

For authentication, I'm planning to support all three methods that django-storages offers: connection strings, account key authentication, and managed identity. This should give users flexibility whether they're doing local development, container deployments, or enterprise Azure setups.

I'll start working on this following the contributor guidelines (CLA, testing, code standards, etc.).

Thanks again for maintaining such a welcoming project for contributors!

[vabene1111]

all of that sounds perfect, thank you for being so considered and well thought trough. Looking forward to your PR.

regarding the cache I think you have already found out everything there is to find out. As long as the URL stays the same in multiple requests the browsers will be able to cache the image without any issues.