ListenBrainz fetch fails on IPv4-only hosts unless HTTP/2 is disabled

[spacecitycoog]

When running Explo in Docker on an IPv4-only host (IPv6 disabled), Explo fails to fetch ListenBrainz playlists and logs errors such as:

• failed to get weekly-exploration playlist, check if ListenBrainz has generated one this week

• failed to make request to ListenBrainz API: EOF

This occurs even though the playlists clearly exist and can be retrieved manually.

Environment

• Host OS: Unraid (Docker)

• Explo image: ghcr.io/lumepart/explo:latest

• Discovery service: ListenBrainz

• Music system: Navidrome (Subsonic API)

• IPv6: disabled on host and inside container
  /proc/sys/net/ipv6/conf/all/disable_ipv6 = 1

• Timezone: correctly matched between ListenBrainz and container

Symptoms

• ./explo --playlist=weekly-exploration fails consistently
• Same failure for weekly-jams and daily-jams
• Logs misleadingly suggest the playlist does not exist
• No authentication or configuration errors are reported

Debugging Findings
Inside the container:

Direct API access works:
curl https://api.listenbrainz.org/1/user//playlists/createdfor
returns valid playlist JSON.

Curl output shows:

• IPv6 connection attempts fail (Network unreachable)
• IPv4 succeeds
• Server negotiates HTTP/2

However, Explo’s internal request fails with an EOF error, suggesting an HTTP/2 client issue when IPv6 is unavailable

Workaround
Force Go to disable HTTP/2 by setting:

environment:
GODEBUG: http2client=0

After applying this:

• Explo immediately succeeds
• Playlists are fetched correctly
• Downloads begin as expected
• No other configuration changes required

Example successful log:
INFO Starting Explo...
INFO initiating search track="Artist - Song"