Instead of trying to match the albums host with current host
(which always fails in path-based routing), use a dummy domain
'albums.localhost.invalid' that will never match the actual host.
This prevents the automatic redirect to /shared-albums.
Version bump to 0.3.2
The app was redirecting to /shared-albums because albumsAppOrigin()
returned the same host as the current URL after runtime replacement.
By setting NEXT_PUBLIC_ENTE_ALBUMS_ENDPOINT at build time to a
placeholder and replacing it at runtime with the full path-based URL,
the host comparison will fail and prevent the redirect loop.
Version bump to 0.3.1
- Fixed admin-helper.sh to use correct Museum binary path (/app/museum-bin/museum)
- Updated start.sh to handle missing S3 configuration gracefully
- App now starts in configuration mode when S3 is not configured
- Shows helpful configuration page instead of failing health checks
- Properly starts Museum server once S3 is configured
- Updated CloudronManifest.json to version 0.2.2
- All web frontends (photos, accounts, auth, cast, albums, family) verified working
- Museum API server running successfully with S3 storage
- Add specific _next asset routes for accounts, auth, cast apps
- Add image asset routes for each app
- Ensure each app's assets are served from correct directory
- Keep photos app routing unchanged
Should fix accounts/auth/cast apps loading issues.
- Change NEXT_PUBLIC_ENTE_ENDPOINT to relative /api for domain flexibility
- Remove runtime JS endpoint replacement (fragile, now unnecessary)
- Fix all domain references to use CLOUDRON_APP_DOMAIN consistently
- Add /ping health check endpoint to Caddy configuration
- Update placeholder server to use dynamic domain
Photos app now working, other apps may need additional fixes.
- Patch origins.ts during Docker build to use window.location.origin + '/api'
- Update version to 0.1.69 to force rebuild
- Add browser compatibility check for server-side rendering
- Fix both API and uploader endpoint redirections
This addresses the root cause where web apps were hardcoded to use
https://api.ente.io instead of the local Museum server.
- Fixed Next.js static asset (_next/*) routing for each app separately
- Updated app path handling to work with both /app and /app/* patterns
- Resolved 404 errors for static assets from auth, accounts, and cast apps
- Updated to version 0.1.66
- Added detailed API endpoint information in SETUP-INSTRUCTIONS.md
- Documented API usage with Ente CLI
- Enhanced routing configuration for auth/cast/accounts apps
- Updated to version 0.1.64
- Change NEXT_PUBLIC_ENTE_ENDPOINT from "/api" to "https://example.com/api" during build to satisfy URL constructor requirements
- Add runtime replacement in start.sh to replace placeholder with actual domain endpoint
- This resolves the "TypeError: Failed to construct 'URL': Invalid URL" error in the frontend
- Changed are_local_buckets from false to true (required for external S3)
- Simplified S3 configuration to only use b2-eu-cen bucket
- Removed unnecessary replication buckets for single bucket setup
This aligns with Ente's documentation where are_local_buckets=true
is used for external S3 services like Wasabi.
- Remove dynamic S3 configuration loading
- Hardcode Wasabi credentials as requested
- Use proper Ente S3 configuration format with datacenter names
- Configure all three storage buckets (b2-eu-cen, wasabi-eu-central-2-v3, scw-eu-fr-v3)
- Set are_local_buckets to false for external S3
- Add compliance flag for Wasabi bucket
This should fix the MissingRegion error by properly configuring S3 storage
according to Ente's expected format.
- Changed Museum server to run on port 8080 instead of 3080
- Updated all health check URLs to use port 8080
- Updated Caddy reverse proxy to forward API requests to port 8080
- Added clarifying comment about port usage
This resolves the circular reference where both Caddy and Museum were trying to use port 3080.
