[ci skip] Fix code review findings: correct Alertmanager URL, add atomic to Loki, remove dead minio NFS export, update design doc

docs/plans/2026-02-13-centralized-log-collection-design.md

@@ -62,7 +62,7 @@ Single-binary deployment. Existing Helm chart config in `loki.yaml`, updated wit
 
 **Ruler:**
 - Evaluates LogQL alert rules in real-time (before chunk flush)
-- Fires to `http://alertmanager.monitoring.svc.cluster.local:9093`
+- Fires to `http://prometheus-alertmanager.monitoring.svc.cluster.local:9093`
 
 **Storage:**
 - NFS PV/PVC at `/mnt/main/loki/loki` (15Gi, existing)
@@ -70,7 +70,7 @@ Single-binary deployment. Existing Helm chart config in `loki.yaml`, updated wit
 
 **Resources:**
 - Memory: 6Gi limit
-- CPU: 500m limit
+- CPU: 1 limit
 
 ### 3. Alloy (Helm Release)
 
@@ -82,7 +82,7 @@ DaemonSet log collector. Existing config in `alloy.yaml` is complete:
 
 **Resources per pod:**
 - Memory: 128Mi limit
-- CPU: 100m limit
+- CPU: 200m limit
 
 ### 4. Grafana Datasource
 

modules/kubernetes/monitoring/loki.tf

@@ -7,6 +7,7 @@ resource "helm_release" "loki" {
   chart      = "loki"
 
   values  = [templatefile("${path.module}/loki.yaml", {})]
+  atomic  = true
   timeout = 300
 
   depends_on = [kubernetes_config_map.loki_alert_rules]

modules/kubernetes/monitoring/loki.yaml

@@ -34,7 +34,7 @@ loki:
       type: local
       local:
         directory: /loki/rules
-    alertmanager_url: http://alertmanager.monitoring.svc.cluster.local:9093
+    alertmanager_url: http://prometheus-alertmanager.monitoring.svc.cluster.local:9093
     ring:
       kvstore:
         store: inmemory