Lotus CIandClaude Opus 5.5
d9f023762e
Lint / JS (eslint) (push) Successful in 12s
Lint / Shell (shellcheck) (push) Successful in 13s
Lint / No secrets in webhook configs (push) Successful in 6s
Lint / Landing page is rendered (matrix (push) Successful in 5s
Lint / Python (ruff) (push) Successful in 6s
Lint / Secret scan (gitleaks) (push) Successful in 8s
Lint / Shell (shellcheck) (pull_request) Successful in 10s
Lint / JS (eslint) (pull_request) Successful in 8s
Lint / No secrets in webhook configs (pull_request) Successful in 5s
Lint / Python deps (pip-audit) (push) Successful in 41s
Lint / Landing page is rendered (matrix (pull_request) Successful in 4s
Lint / Python (ruff) (pull_request) Successful in 5s
Lint / Secret scan (gitleaks) (pull_request) Successful in 11s
Lint / Python deps (pip-audit) (pull_request) Successful in 56s
livekit-graceful-restart.sh decided "no active calls" by counting established TCP connections on :7881 with `ss ... state established | grep -c ESTAB`. Two bugs: - With a `state` filter ss omits the State column, so `grep -c ESTAB` is always 0: the check never saw anything. - :7881 is only LiveKit's ICE-TCP media fallback. Signalling is on :7880 and media is normally UDP. Reproduced in a two-person call on the local stack: the old check counted 0 and would have restarted. Now it waits while LiveKit's own livekit_participant_total is above 0 or any signalling WebSocket on :7880 is open, and also waits if the metrics can't be read. The deploy hook now installs the script when it changes (it never did, so edits to it never reached the server). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_017PCb59uY3UKqDqt7B6hzd5
36 lines
1.6 KiB
Bash
36 lines
1.6 KiB
Bash
#!/bin/bash
|
|
# Checks every 5 minutes if a livekit restart is pending.
|
|
# Only restarts when LiveKit reports no participants AND nobody holds a
|
|
# signalling connection, so a restart never drops a call.
|
|
# Run by: livekit-graceful-restart.timer (systemd)
|
|
|
|
if [ ! -f /run/livekit-restart-pending ]; then
|
|
exit 0
|
|
fi
|
|
|
|
# Participants LiveKit itself counts, across all rooms. Counting TCP connections
|
|
# on :7881 (the old check) only saw clients on the ICE-TCP fallback: signalling
|
|
# is on :7880 and media is normally UDP, so most calls looked idle.
|
|
METRICS_URL="${LIVEKIT_METRICS_URL:-http://127.0.0.1:6789/metrics}"
|
|
if ! METRICS=$(curl -sf --max-time 5 "$METRICS_URL") \
|
|
|| ! printf '%s\n' "$METRICS" | grep -q '^livekit_participant_total'; then
|
|
echo "$(date): Livekit restart pending but its metrics are unreadable — assuming a call is active, waiting."
|
|
exit 0
|
|
fi
|
|
PARTICIPANTS=$(printf '%s\n' "$METRICS" | awk '/^livekit_participant_total/ {s += $NF} END {printf "%d", s}')
|
|
|
|
# Belt and braces: open signalling WebSockets (a client mid-join may not be a
|
|
# participant yet). Count lines (-H drops the header): with a `state` filter ss
|
|
# omits the State column, so the old `grep -c ESTAB` was always 0.
|
|
SIGNALLING=$(ss -Htn state established '( sport = :7880 )' | wc -l)
|
|
|
|
if [ "$PARTICIPANTS" -gt 0 ] || [ "$SIGNALLING" -gt 0 ]; then
|
|
echo "$(date): Livekit restart pending but $PARTICIPANTS participant(s), $SIGNALLING signalling connection(s) — waiting."
|
|
exit 0
|
|
fi
|
|
|
|
echo "$(date): No active calls — applying livekit-server restart."
|
|
systemctl restart livekit-server
|
|
rm -f /run/livekit-restart-pending
|
|
echo "$(date): livekit-server restarted successfully."
|