The rollup and the backup check in once, when they end, with the run's duration

A start and a finish sent seconds apart can reach Sentry in either order,
and a start that lands second is never closed, so a fast rollup could be
reported as a timeout on a night it succeeded. Hnefatafl's monitor did so
nine nights in fifteen on the shared host. One check-in at the end cannot
race; a run that hangs sends nothing and is marked missed. Installed on
the host by hand; the next deploy installs the same files.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GwFKMuQnPAEHJ5yA1q4orh
This commit is contained in:
Eric Wagoner
2026-10-10 17:50:55 -04:00
co-authored by Claude Opus 5.5
parent aeb58bdf4a
commit ad23c83b8a
2 changed files with 12 additions and 4 deletions
+6 -2
View File
@@ -11,7 +11,12 @@ STAMP=$(date +%Y-%m-%d)
# Sentry Crons check-in: /root/.wizwar-sentry-cron holds the check-in URL
# (auto-creates the monitor on first ping). Absent file = no check-ins.
CRON_URL=$(cat /root/.wizwar-sentry-cron 2>/dev/null || true)
checkin() { [ -n "$CRON_URL" ] && curl -sf -o /dev/null "$CRON_URL?status=$1" || true; }
# One check-in, when the run ends, carrying its duration: a start and a
# finish sent seconds apart can reach Sentry in either order, and a start
# that lands second is never closed. A run that never ends sends nothing,
# and Sentry marks the night missed.
START=$SECONDS
checkin() { [ -n "$CRON_URL" ] && curl -sf -o /dev/null "$CRON_URL?status=$1&duration=$((SECONDS - START))" || true; }
if grep -q CHANGE_ME /root/.config/rclone/rclone.conf; then
echo "$(date -Is) skipped: Spaces credentials not configured yet" >> "$LOG"
@@ -20,7 +25,6 @@ fi
{
echo "=== $(date -Is) backup start"
checkin in_progress
if rclone sync "$SRC" "spaces:$BUCKET/wizwar/current" 2>&1 &&
rclone copy "$SRC" "spaces:$BUCKET/wizwar/snapshots/$STAMP" 2>&1; then
checkin ok
+6 -2
View File
@@ -21,13 +21,17 @@ CRON_URL=$(cat /root/.wizwar-sentry-cron-rollup 2>/dev/null || true)
# The check-in carries the schedule: Sentry only keeps a monitor it has
# been told the shape of, and a bare ping to an unknown slug is accepted
# and dropped without a word.
# One check-in, when the run ends, carrying its duration: a start and a
# finish sent seconds apart can reach Sentry in either order, and a start
# that lands second is never closed. A run that never ends sends nothing,
# and Sentry marks the night missed.
START=$SECONDS
checkin() {
[ -n "$CRON_URL" ] && curl -sf -o /dev/null -X POST -H "Content-Type: application/json" \
-d "{\"status\":\"$1\",\"monitor_config\":{\"schedule\":{\"type\":\"crontab\",\"value\":\"$SCHEDULE\"},\"checkin_margin\":30,\"max_runtime\":10,\"timezone\":\"UTC\"}}" \
-d "{\"status\":\"$1\",\"duration\":$((SECONDS - START)),\"monitor_config\":{\"schedule\":{\"type\":\"crontab\",\"value\":\"$SCHEDULE\"},\"checkin_margin\":30,\"max_runtime\":10,\"timezone\":\"UTC\"}}" \
"$CRON_URL" || true
}
checkin in_progress
if python3 - "$DAY" "$OUT" <<'PY' >> "$LOG" 2>&1
import json, sys, os, glob, gzip, datetime, collections, subprocess, re
day, out = sys.argv[1], sys.argv[2]