See CHANGELOG.md for details.
Fixes
- Self-hosted priority endpoints (Tidal/Qobuz/Amazon) were configured but never actually used (#13); a self-hosted instance sat there doing nothing while every request still hit the public pool, which is what was actually getting rate-limited, and the source got reported as fully dead even though your own instance was fine. Self-hosted endpoints are now tried first, public pool only as fallback if self-host is actually down.
- Status page could show a source as "dead" while its self-host instance was healthy; fixed.
- A source with no valid endpoints configured (e.g. a rejected self-host URL) was misreported as "upstream is down" instead of "nothing configured"; mostly affected Amazon.
- Double
soulseek: soulseek:prefix in fallback error messages, and a misattribution edge case.
Known limitation: a self-hosted endpoint that hangs (accepts the connection, never responds, no error) isn't detected as unhealthy yet; real failures (5xx/401/403/refused/DNS) are handled correctly.
Internal
flacidal-corebumped to v0.20.0; that's where the actual endpoint-pool rework lives.
Full Changelog: v4.16.0...v4.17.0