fix: address Copilot review comments with KV-safe periodic save

- Add 15-min periodic save (isPeriodicSave) so latency/history stay fresh
  without writing every 5-min tick (~96 writes/day vs ~288)
- Track _lastIssuesFetchAttemptDay to bound daily retries to once per day
  regardless of success/failure; log non-2xx HTTP status
- Update _lastSaveTime on every save so the 15-min clock resets correctly
- Clarify wrangler.toml cron comment: change was intentional to reduce writes

Agent-Logs-Url: https://github.com/SillyLittleTech/UpMagic/sessions/9db4878c-bad8-4268-8318-8280f3ca1f11

Co-authored-by: kiyarose <75678535+kiyarose@users.noreply.github.com>

Author: Copilot

worker/index.js

@@ -128,6 +128,10 @@ async function runHeartbeat(env) {
   const heartbeatAt = Date.now();
   const todayUTC = new Date(heartbeatAt).toISOString().slice(0, 10); // "YYYY-MM-DD"
 
+  // Save at most every 15 minutes for latency/history freshness (~96 writes/day on free tier)
+  const PERIODIC_SAVE_INTERVAL_MS = 15 * 60 * 1000;
+  const isPeriodicSave = heartbeatAt - (state._lastSaveTime || 0) >= PERIODIC_SAVE_INTERVAL_MS;
+
   for (const target of targets) {
     if (state.services[target.id]?.isManual) {
       continue; 
@@ -175,9 +179,13 @@ async function runHeartbeat(env) {
     console.log(`[heartbeat] ${target.name}: ${newStatus}${latency !== null ? ` (${latency}ms)` : ''}`);
   }
 
-  // Fetch maintained issues from ProjectHub at most once per day
-  const lastIssuesFetchDate = state._lastIssuesFetchDay || '';
-  if (lastIssuesFetchDate !== todayUTC) {
+  // Fetch maintained issues from ProjectHub at most once per day.
+  // Track attempt day separately from success day so a non-2xx response
+  // doesn't cause unbounded retries on every heartbeat.
+  const lastIssuesFetchAttempt = state._lastIssuesFetchAttemptDay || '';
+  if (lastIssuesFetchAttempt !== todayUTC) {
+    state._lastIssuesFetchAttemptDay = todayUTC; // record attempt regardless of outcome
+    needsSave = true; // persist the attempt date so retries are bounded
     try {
       const issuesRes = await fetch('https://api.github.com/repos/SillyLittleTech/projecthub/issues?labels=maintain&state=open', {
         headers: {
@@ -190,17 +198,23 @@ async function runHeartbeat(env) {
         state.maintenance = issues.map(i => ({ title: i.title, url: i.html_url, created_at: i.created_at }));
         state._lastIssuesFetchDay = todayUTC;
         console.log(`[daily] Fetched ${issues.length} maintenance issue(s) for ${todayUTC}`);
-        needsSave = true; // persist updated maintenance list
+      } else {
+        console.error(`[daily] Issues fetch failed: HTTP ${issuesRes.status} for ${todayUTC}`);
       }
     } catch (e) {
       console.error('Failed to fetch maintenance issues', e);
     }
   }
 
-  // Only write to KV when something changed, or once per day as a daily summary
+  // Only write to KV when something meaningful changed:
+  //   - status change: immediate (for accurate frontend display)
+  //   - periodic (every 15 min): keeps latency/history reasonably fresh (~96 writes/day)
+  //   - daily: ensures a flush even during quiet periods + logs a summary
+  //   - needsSave: issues fetch attempt date or maintenance list must be persisted
   const lastDailySaveDate = state._lastDailySaveDay || '';
   const isDailySave = lastDailySaveDate !== todayUTC;
-  if (statusChanged || needsSave || isDailySave) {
+  if (statusChanged || needsSave || isDailySave || isPeriodicSave) {
+    state._lastSaveTime = heartbeatAt; // reset the 15-min clock on any save
     if (isDailySave) {
       state._lastDailySaveDay = todayUTC;
       const summary = Object.entries(state.services)

wrangler.toml

@@ -17,4 +17,4 @@ id = "d6179b2f901645de81b9805c4e7ac591"
 preview_id = "59e7164e44154d7fa79b2a093b338530"
 
 [triggers]
-crons = ["*/5 * * * *"] # Run every 5 minutes for heartbeat
+crons = ["*/5 * * * *"] # Every 5 minutes (intentionally relaxed from 1 min to cut KV writes ~5x)