Help understanding pathway of data for multi-user maloja + navidrome setup

[jessienab]

I have two users, myself and my friend, using my single Navidrome music instance.
We each have our own separate account logins.

We each have a maloja instance; mine has been used for a while, whereas my friend's is only now getting new data added (previously was just historical reference).

I am wondering if I'm understanding how to setup MS properly:

1. Setup two scrobble sources with MS under Subsonic setup, one for me, one for friend.
2. Setup two scrobble destinations with MS under Maloja setup, one for my instance, one for my friend's.
3. Point my scrobble source to my scrobble destination (my Maloja instance)
4. Do the same for my friend's instance.

With this setup, scrobbles are handled by MS, therefore I need to turn off scrobble integration with Navidrome, is that all correct? And MS will look for actively played/playing songs.

And therefore after, MS will properly route what I play only to my Maloja instance, and only what my friend plays to their instance (instead of Navidrome only sending data to a single configurable scrobble instance).

Please let me know if this seems correct, just want to understand it before diving in with config.
Thanks!

[FoxxMD]

Yes, your understanding of the setup is correct!

My only suggestion would be to keep the scrobbling integration with Navidrome and then use two Listenbrainz (Endpoint) Sources, instead of Subsonic sources. The "Scrobble to ListenBrainz" setting, in navidrome, is a user-specific setting so you should be able to configure different tokens for each user, which MS can then use to differentiate which Source is receiving the scrobble.

The legacy "now playing" endpoint for navidrome is not great for monitoring realtime activity, the integrated scrobbling "scrobble to listenbrainz" is much more accurate than it. Navidrome has just released a new version that includes much more accurate now playing reporting but I have not yet implemented this into the subsonic source. So for now you're better off using the listenbrainz endpoint source.