Player
- Arrow keys seek by 5s, Shift for a 30s jump on long recordings, J/L for +-30s,
Space toggles playback, and up/down adjust volume. Dragging a scrubber across a
multi-hour replay is unusable — one pixel is tens of seconds — so keys are the
practical way to move around a long recording.
- The shortcuts ignore key events from inputs, so typing a chapter time in the
replay editor no longer scrubs the video.
- Add a playback speed selector. It is a replay tool: speeding a live stream up
would only push it further ahead of the encoder's buffer.
Replay editor
- Stop playback restarting on every edit. The player's source effect depended on
the `lines` and `qualities` arrays, and callers such as the replay editor pass
fresh literals on every render, so adding a chapter or typing a title was
treated as a source change and reloaded the stream from the beginning. The
effect now depends on the resolved source URL instead of the arrays it came
from. Measured: adding three chapters leaves the media requests unchanged and
the playhead advancing.
Settings persistence
- Settings edited in the UI were applied to the in-memory settings object and
never written down, so the brand, transcode ladder and load-balancing strategy
silently reverted to the config.py defaults on every restart. They are now
saved to backend/settings.json (atomic temp-file replace) and re-applied at
startup, with only known keys accepted so a hand-edited file cannot inject
arbitrary attributes.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Replay
- Hand the replay file to nginx with X-Accel-Redirect instead of streaming it
through the backend. A replay is tens of gigabytes, so Python was copying the
whole file and holding a worker for the entire transfer: measured 8 MB in
8.46s (~1 MB/s). nginx serves it from disk with sendfile at ~400 MB/s (1 GB in
2.68s) and the backend process does not appear in CPU usage during the transfer.
Authentication and the path-traversal check still run in Python; only the bytes
moved. The internal location is what keeps the path from being reachable
directly and bypassing that check.
- nginx reloads and the media directory is aliased internally.
Settings
- GET /settings/transcode-profiles returned the bare mapping while the settings
page reads `data.profiles`, so it always saw undefined, fell back to an empty
object, and every save posted `{"profiles": {}}` — which the PUT rejects as
"at least one profile". Saving settings could never succeed. The response now
matches the shape PUT accepts, verified by round-tripping it.
Auth
- Derive the admin token instead of storing it. The token is
sha256(password + ":token"), but it was kept in a module-level set, which only
works with one process: with uvicorn --workers N, logging in on one worker left
the others rejecting the token and the admin appeared logged out at random.
Viewer
- Show the real session id in the stats panel; the previous code assigned it
from a local variable that was never populated, so it always read "不可用".
- Send the first heartbeat before the first command poll. Both fired together on
mount, so the poll reached the backend before the session existed and 404'd on
every page load.
Housekeeping
- Ignore backend/nodes.json; the admin node manager writes it at runtime.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Hardware encoding
- Detect the best usable encoder at startup by encoding a test frame with each
candidate in turn (nvenc, qsv, amf, then libx264). Being compiled into ffmpeg
is not enough: h264_amf fails without amfrt64.dll and h264_qsv fails without an
Intel iGPU, so the probe is the only reliable answer.
- Translate the quality target into the chosen encoder's own vocabulary. libx264's
-crf and "ultrafast" mean nothing to h264_nvenc, which takes -cq and p1..p7, so
passing the configured values straight through would have been silently wrong.
- On the reference GTX 1060 a 1080p->720p rendition costs ~62% of a core on NVENC
versus ~109% on libx264 ultrafast.
Per-rendition supervision
- Track encoder processes per rendition, not per stream. is_running answered
"any alive", so when the source ("原画") encoder died while 1080p and 720p kept
running, the supervisor saw a healthy stream and never restarted it — the
source URL then 404'd permanently. It now requires every rendition to be alive
and reports which ones are down.
- Restart only the dead renditions, so a single failure no longer interrupts
viewers on the renditions that were fine.
Production serving
- Serve the built frontend from FastAPI with SPA fallback, so production is one
process and one port with no Node runtime. Node becomes a build-time dependency.
- start.ps1/start.sh build the frontend on first run instead of failing to a 503.
Cleanup
- Drop the duplicated top-left LIVE badge; the control strip already carries it.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Frontend
- Replace the Ant Design app with the shadcn/ui app as the single frontend,
keeping the portfolio side-by-side revertible in this commit.
- Drop the abandoned mdui experiment and the AntD tree so there is one app,
one router and one design system to maintain.
Live playback
- Stop racing the encoder. lowLatencyMode + liveSyncDurationCount:1 targeted a
single 4s segment, so playback ran at up to 1.10x and the forward buffer
bottomed out at 0.1s, re-buffering every segment. Measured over 40s:
buffer min 0.1s -> 8.9s, rate 1.00 constant, 0 stalls.
- Remove the 落后 labels and the page-level edge-snapping timer that fought
hls.js over two different definitions of "latency".
- Live can no longer be paused: the control is not rendered while it plays, so
the icon can never misreport state.
- Fix autoplay: muted autoplay started silently and never unmuted, and the
gesture that restored sound also toggled playback off. One owner now.
Source ("原画") rendition
- Publish the publisher's stream remuxed losslessly via -c:v copy alongside the
transcoded ones, so a source option exists even when every ladder entry
downsizes.
- Give renditions an ASCII path slug: the HLS path is part of the URL and
MediaMTX rejects non-alphanumeric paths, so 原画 could never be served.
Backend
- Supervise MediaMTX from the backend: start it at boot and restart it on exit,
never killing an origin it did not start.
- Clear stale HLS output before restarting an encoder; an ENDLIST-less playlist
left by a dead encoder hangs hls.js forever.
- Mark JSON timestamps as UTC. utcnow().isoformat() emitted a naive string that
JS read as local time, so every viewer heartbeat looked 8 hours old in UTC+8
and the live board's active-viewer count was permanently zero.
- Rebind a viewer session when it arrives on another stream instead of failing,
and survive the StrictMode double-heartbeat race on first insert.
- Replace the Linux-only /proc and os.getloadavg metrics with psutil.
- Complete MediaMTX's cookieCheck redirect inside /hls proxying, and hold the
httpx client open for the streamed body.
Housekeeping
- Ignore RainOSS (it stores accessKeySecret), runtime output, TLS keys and
.venv; remove generated screenshots, load test scripts and stray files.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Backend: FastAPI + SQLAlchemy + SQLite
Frontend: React + Vite + Ant Design (serif Anthropic theme)
Media server: MediaMTX with edge node support
Load balancing: round-robin / weighted / least-conn
Multicast detection with automatic fallback
Transcoding support via FFmpeg (configurable quality)
hls.js player with line & quality switching
Admin panel with password auth
Windows compatible scripts
