Skip to content

v4.17.0

Choose a tag to compare

View all tags
@github-actions github-actions released this 27 Aug 03:47
· 41 commits to main since this release
v4.17.0
40bdc17

Fixes

  • Self-hosted priority endpoints (Tidal/Qobuz/Amazon) were configured but never actually used (#13): the setting existed in Settings and got saved, but nothing in the desktop app or the headless server (cmd/server) ever wired it into the endpoint pool. A self-hosted instance just sat there doing nothing: every request still went through the public community pool, which is what was actually getting rate-limited and blacklisted, and the source got reported as fully dead even though your own instance was fine the whole time. Root cause lived in flacidal-core (see flacidal-core's changelog for the pool internals). Self-hosted endpoints are now tried first, with the public pool only as fallback if self-host is actually down.
  • Status page could still show a source as "dead" while its self-host instance was healthy: a blacklisted-but-not-dead endpoint counted as "up" for the self-host badge but "down" for the overall status, so the two could contradict each other on the same row. The overall status now defers to self-host health when it's the one keeping a source alive.
  • A source with no valid endpoints configured at all (e.g. a rejected self-host URL) was misreported as "upstream is down" instead of "nothing configured"; mattered most for Amazon, whose pool has no fallback-to-defaults if every configured URL gets rejected by the security filter (must be https://, or http:// on loopback/private addresses only).
  • Double soulseek: soulseek: prefix in fallback error messages, and a couple of cases where a Soulseek failure could get misattributed.

Known limitation

A self-hosted endpoint that accepts a connection and then never responds (hangs, no error) isn't currently detected as unhealthy; only real failures (5xx, 401/403, connection refused, DNS failure) are. If your self-host instance is up but broken in a way that just hangs requests, it'll keep getting tried and slow things down rather than falling back. A real crash or an HTTP error response is handled correctly.

Internal

  • Bumped flacidal-core to v0.20.0: see its changelog for the pool/tier internals.

Full Changelog: v4.16.0...v4.17.0

Assets 5
Loading