feat: A/V health monitor, buffer bar and stream health dashboard
Backend (restream.py):
- Add _health_check_loop task: checks data flow every 5 s, updates
av_health (ok/warning/error), auto-corrects stuck ffmpeg via SIGTERM
- Health thresholds: warning >12 s, error >25 s — normal reconnects
(1-4 s) no longer trigger false warnings
- Count ffmpeg stderr A/V-warning events (av_desync_count) as metric
- Expose buffer_pct (server-side queue fill of slowest client),
corrections (health-triggered ffmpeg restarts), av_health in stats()
- Revert wallclock timestamp flag: -use_wallclock_as_timestamps 1 was
generating hundreds of spurious demuxer warnings per session, causing
ffmpeg to stall and producing continuous video freezes
- Simplify ffmpeg cmd to minimal safe set: -reconnect 1
-reconnect_streamed 1 -reconnect_delay_max 2
Dashboard (Dashboard.tsx):
- Add 6th StatCard "Salud A/V" with dynamic color and correction count
- Add "Salud A/V" column per stream: badge ok/aviso/error,
corrections counter (violet), desync event count (grey)
- Add BufferBar: thin progress bar showing server queue fill %
(green <40 %, yellow 40-70 %, red >70 %)
- Alert banner in header when any stream is in error av_health state
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
joaquin
57c8f6fb64