Funkwhale on a RPi 1 A


#1

I’ve deployed FW on a first-gen raspberry pi (2013), with around 30k tracks stored on a remote NAS, and since the documentation is asking for feedback about there, here I am.

  • Loading 50 artists via http://192.168.2.49/library/artists?query=&page=1&paginateBy=50&ordering=name is taking 24 seconds.
  • Search via http://192.168.2.49/library/artists?query=within%20&page=1&paginateBy=50&ordering=name is taking 9 seconds.
  • The scan for all my music ran overnight, so no idea about how much time it took
  • Accessing the library page via http://192.168.2.49/library takes 1 minutes, which is weird because I don’t have any favourite, nor playlist, nor even tried to listen to music
  • The music streaming is working, altough I’ve disabled transcoding because I don’t need it.

So, in conclusion, it’s almost useable.

What other metrics could you be interested in?


#2

Hi @jvo ! Thanks for doing this experiment :slight_smile:

What kind of install did you choose (docker ? all in one ? bare metal ?) ?


#3

@jvo thank you so much for the feedback, it’s really appreciated. Response time greater than a few seconds affect the usability of the app, and I’d like to improve those if possible.

Usually, the bottolneck in web applications is the time needed to contact the database and run SQL queries, but this may not be the issue here. One way to detect that would be to backup your database, visit /api/admin/music/artist, delete everything, import only a single track, then try to load various pages through the app and see how it affect the response times.

(You can, of course, restore your database backup after that :wink:

If response time improve after that, this probably means we could add some indexes in the database or optimize the way we query it.


#4

I deployed it using the barebone debian method.

I’ll try to test the database-thingy this week. A quick look via the web-inspector indicates that the request that is taking the most time is indeed the one to the API to get the artists.


#5

That’s good news, because it’s the most flexible method to investigate this kind of performance issue!

Thank you for your time :slight_smile:


#6

My SD card died, and I’m too lazy to redeploy an unsaleable funkwhale instance from scratch :confused:

But I guess that optimizing database queries and using indexes is still a good idea :wink:


closed #7

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.