The previous -max_alloc approach does not fix the 'overlong headers' error
because FFmpeg 5.1 (Debian bookworm) uses a compile-time stack-allocated
buffer (MAX_URL_SIZE=4096) for HTTP response headers, which -max_alloc
cannot change.
Instead, when the stream URL is HTTP/HTTPS, launch curl to handle the
HTTP connection and pipe its stdout to FFmpeg via stdin (pipe:0). Curl
does not have the 4096-byte header limit, so it handles oversized CDN
response headers from Douyin without error.
Additional changes:
- RequestStopAsync kills curl first (instead of sending 'q' to FFmpeg),
causing the pipe to close and FFmpeg to exit gracefully on EOF.
- SessionProcessRuntime tracks the curl process for cleanup.
- Add -max_alloc 100000000 to FFmpeg arguments for HTTP inputs to avoid
'overlong headers' error when CDN (e.g. Douyin) returns oversized HTTP
response headers exceeding FFmpeg's default 4096-byte buffer.
- Add exponential backoff for repeated startup failures (30s → 15min cap)
to break the tight fail→retry→re-poll loop that floods notifications.
- Throttle startup failure notifications to at most one per 30 minutes per
room to prevent email/webhook storms during persistent failures.
- Reset backoff counter when a session successfully opens its first segment.
Under the Red storage tier, MP4 finalization (TS->MP4 remux) was being skipped, so tasks never reached Completed and the segment_completed event script — which uploads the file and deletes the local source to free space — never ran. The disk could never recover, deadlocking all recording and transcoding.
Two reversed checks caused this: (1) FfmpegService gated finalization on the legacy HasEnoughSpace MB threshold (effectively 4GB) instead of the tier system, and (2) the polling loop only resumed paused finalizations when NOT in the Red tier. Now finalization is gated solely on ShouldPauseActive (true Red only) and the polling loop always attempts to resume it every cycle, since finalization is the very mechanism that frees space. Once any segment finalizes, the upload+delete script runs and the disk recovers, letting the rest finish.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
nanxun
