Skip to content

NAS recorder

github-actions[bot] edited this page Nov 22, 2026 · 7 revisions

The recorder is clippa record watch in a Docker container on the NAS: it records each show live, with chat, into the streams share, and the worker takes each recording from there.

Warning

Deploy only when no show is live: a restart starts a new recording part.

Deploy

bash deploy/record/deploy.sh from the repo root (ssh alias nas). The recorder reads [record] and [kick] from the clipper.toml in the NAS folder, the one copy since 2026-11-05 (NAS hub): the script copies ./clipper.toml there only if there is none (or CLIPPA_TOML=<file> to replace it). Re-run it after changing [record] channels. It refuses while a show is being recorded (the restart would start a new recording part) and when the recorder's state can't be read: clippa record status shows what the recorder is doing, from the heartbeat it writes to the share (exit 3 while live, 1 when the heartbeat is unreadable), and deploy.sh --force deploys anyway. clippa services stop|restart recorder stops or restarts it over ssh without a new build, with the same check.

Cut on the NAS (#51)

The same image cuts segments and compilations beside the source, instead of the laptop pulling and pushing every byte over the share (it writes at ~23 MB/s on Wi-Fi). In clipper.toml: [cut] backend = "nas", nas_host (the ssh alias, key login), nas_data (the recorder's CLIPPA_DATA, e.g. /volume1/docker/clipper) and nas_user (its PUID:PGID, e.g. 1000:10). Each clippa cut then sends this checkout's clippa/ package and clipper.toml over ssh and runs in a one-off container (clipper-cut-<stream>), so the recorder needs no redeploy for code changes, only for a new image. Picks still cut here (they re-encode). clippa doctor --role gpu checks the ssh login and the image. A cut stopped half way can leave its container running: ssh nas docker rm -f clipper-cut-<id>.

Clone this wiki locally