Errors during brainzSource polling.

[jacksaturn]

I've noticed that there are regular errors in my logs related to polling ListenBrainz as a source (brainzSource). This is causing brainzSource to go into an idle state. It does seem to periodically emerge from the idle state to grab data, but I'm wondering if there is anything I can do to avoid this happening.

Is it possible I've got something misconfigured, or that I'm polling brainzSource too often? Or are these just connection errors on the ListenBrainz server?

I'm attaching some WARN and ERROR logs from the last hour as an example.

JS-MS-error_logs.txt

[FoxxMD]

This is, unfortunately, the cost of doing business with Listenbrainz. Anecdotally, their API is down a lot. If not fully "offline" then it will constantly just hang until MS decides to time it out (about 30 seconds of waiting). Sometimes for minutes at a time, then be ok for an hour, then 10's of minutes of timeouts again.

If multi-scrobbler stops polling due to these errors it should try to restart polling every 20 minutes, at which point it should catch up on any missed scrobbles. You can also increase the number of polling retry attempts by using a file/aio config for listenbrainz by adding maxPollRetries to options like this (for listenbrainz.json file):

[
  {
    "name": "brainzSource",
    "enable": true,
    "configureAs": "source",
    "data": {
      "token": "029b081ba-9156-4pe7-88e5-3be671f5ea2b",
      "username": "FoxxMD",
    },
    "options: {
      "maxPollRetries": 5
    }
  }
]

But this is only helpful if their API actually becomes available again in a short amount of time. If its down/timing out for 10s of minutes the retries won't help.

[jacksaturn]

Thanks for the confirmation — that was my hunch, that it was an issue on the ListenBrainz side.

Here's my use case for myLz:

In my custom scrobbling script, I have it taking out all the extraneous version/deluxe/remastered/etc. info in the song and album titles. I found that this resulted in cleaner info on Last.fm. ListenBrainz has always been good about dealing with metadata, so it kept on working fine. However, Koito seemed confused by it, and often when it receives a scrobble title without that extra info, it is saving the tracks internally with the album title matching the song title. So I switched Koito to use ListenBrainz as its source, so that it'll have that added layer of metadata cleaning.

With all of that said, I read a comment on the Koito Github page mentioning a plan to have MS do metadata cleanup / checking via the ListenBrainz database: #363

Once that's implemented perhaps I'll no longer need to use myLz as a go-between. (For the record, all my scrobbles to ListenBrainz seem to be getting there without noticeable error.)