Fix SQL bucketing using Julian Day calculation

server/index.js

@@ -472,19 +472,21 @@ app.get('/api/history/overview', (req, res) => {
       const intervalHours = days <= 7 ? 1 : days <= 14 ? 3 : 6;
       console.log(`Bucketing with ${intervalHours} hour intervals`);
       
-      // Simplified query - directly group by time buckets
-      const intervalSeconds = intervalHours * 3600;
+      // Use a reliable bucketing approach with Julian Day Numbers
+      // Convert to julian day, multiply by 24 for hours, divide by intervalHours and round
+      const bucketCalc = `ROUND((julianday(checked_at) * 24) / ?) * ? / 24`;
+      
       recentChecks = db.prepare(`
         SELECT 
-          datetime((CAST(strftime('%s', checked_at) AS INTEGER) / ?) * ?, 'unixepoch') AS checked_at,
+          datetime(ROUND((julianday(checked_at) * 24) / ?) * ? / 24) AS checked_at,
           AVG(CASE WHEN status='up' THEN response_time ELSE NULL END) AS avg_up_response,
           SUM(CASE WHEN status='up' THEN 1 ELSE 0 END) AS up_count,
           COUNT(*) AS total_count
         FROM checks
         WHERE resource_id = ? AND checked_at > datetime('now', ?)
-        GROUP BY (CAST(strftime('%s', checked_at) AS INTEGER) / ?)
+        GROUP BY ROUND((julianday(checked_at) * 24) / ?) * ? / 24
         ORDER BY checked_at ASC
-      `).all(intervalSeconds, intervalSeconds, resource.id, `-${days} days`, intervalSeconds).map(row => ({
+      `).all(intervalHours, intervalHours, intervalHours, intervalHours, resource.id, `-${days} days`, intervalHours, intervalHours).map(row => ({
         status: row.up_count >= Math.ceil(row.total_count/2) ? 'up' : 'down',
         response_time: Math.round(row.avg_up_response || 0),
         checked_at: row.checked_at,