修复BUG

server/prometheus-service.js

@@ -172,7 +172,8 @@ async function getOverviewMetrics(url, sourceName) {
     // Total traffic transmitted in last 24h
     query(url, 'sum by (instance, job) (increase(node_network_transmit_bytes_total{device!~"lo|veth.*|docker.*|br-.*"}[24h]))').catch(() => []),
     // Up instances (at least one successful scrape in last 5m)
-    query(url, 'max_over_time(up{job=~".*node.*|.*exporter.*"}[5m])').catch(() => [])
+    // We broaden the job filter to catch more variations of node-exporter jobs
+    query(url, 'max_over_time(up{job=~".*node.*|.*exporter.*|.*host.*"}[5m])').catch(() => [])
   ]);
 
   // Build per-instance data map
@@ -252,6 +253,14 @@ async function getOverviewMetrics(url, sourceName) {
     inst.netTx = parseFloat(r.value[1]) || 0;
   }
 
+  // Final check: If an instance has non-zero CPU or Memory total data but is marked offline, 
+  // it means we missed its 'up' metric due to job labels, but it's clearly sending data.
+  for (const inst of instances.values()) {
+    if (!inst.up && (inst.cpuPercent > 0 || inst.memTotal > 0)) {
+      inst.up = true;
+    }
+  }
+
   // Aggregate
   let totalCpuUsed = 0, totalCpuCores = 0;
   let totalMemUsed = 0, totalMemTotal = 0;
@@ -408,7 +417,7 @@ function mergeCpuHistories(histories) {
 async function getTrafficHistoryRange(url) {
   const now = Math.floor(Date.now() / 1000);
   const start = now - 86400; // 24h ago
-  const step = 300; // 5 minutes
+  const step = 5; // 5 seconds (17,280 points for 24h)
 
   const queries = [
     'sum(node_network_receive_bytes_total{device!~"lo|veth.*|docker.*|br-.*"})',