PERF(observability): add HA for Loki and Tempo

- Loki: replicas 1→2 with soft anti-affinity
- Tempo: replicas 1→2 with soft anti-affinity
- Thanos/Prometheus: keep replica 1

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

loki/helm-values.yaml

@@ -45,10 +45,20 @@ loki:
 deploymentMode: SingleBinary
 
 singleBinary:
-  replicas: 1
+  replicas: 2
   # Disable PVC - use emptyDir for WAL (data stored in MinIO)
   persistence:
     enabled: false
+  # Soft anti-affinity for HA
+  affinity:
+    podAntiAffinity:
+      preferredDuringSchedulingIgnoredDuringExecution:
+        - weight: 100
+          podAffinityTerm:
+            labelSelector:
+              matchLabels:
+                app.kubernetes.io/name: loki
+            topologyKey: kubernetes.io/hostname
   extraVolumes:
     - name: data
       emptyDir:

prometheus/helm-values.yaml

@@ -70,7 +70,7 @@ prometheus:
     # Enable remote write receiver for OTel Collector
     enableRemoteWriteReceiver: true
 
-    # Single replica (HA removed for resource optimization)
+    # Single replica
     replicas: 1
     replicaExternalLabelName: prometheus_replica
 

tempo/helm-values.yaml

@@ -10,6 +10,20 @@
 # Medium priority for observability
 priorityClassName: medium-priority
 
+# HA with 2 replicas
+replicas: 2
+
+# Soft anti-affinity for HA
+affinity:
+  podAntiAffinity:
+    preferredDuringSchedulingIgnoredDuringExecution:
+      - weight: 100
+        podAffinityTerm:
+          labelSelector:
+            matchLabels:
+              app.kubernetes.io/name: tempo
+          topologyKey: kubernetes.io/hostname
+
 # =============================================================================
 # Resource Limits (optimized for small cluster)
 # =============================================================================