fix: set ndots:1 on grafana pod dnsConfig

platform/apps/kube-prometheus-stack/values.yaml

@@ -10,6 +10,17 @@ grafana:
   # this one credential.
   plugins:
     - yesoreyeram-infinity-datasource
+  # Not a WAN-latency problem — default ndots:5 makes every external FQDN
+  # lookup (grafana.com here) burn through 4 failed cluster-search-domain
+  # queries before trying the bare name, costing several seconds. That
+  # alone exceeded grafana-cli's request timeout on plugin install and
+  # crash-looped the pod (confirmed: 4-7s hang with ndots:5 vs 8ms with
+  # ndots:1 for the exact same lookup, from the same node). Same root cause
+  # already hit and fixed this way in homelab-services' ignis chart.
+  dnsConfig:
+    options:
+      - name: ndots
+        value: "1"
 
 prometheus:
   prometheusSpec: