diff --git a/k8s/security/ci-cd/forgejo-values.yaml b/k8s/security/ci-cd/forgejo-values.yaml
index 63ebb13..ddaffde 100644
--- a/k8s/security/ci-cd/forgejo-values.yaml
+++ b/k8s/security/ci-cd/forgejo-values.yaml
@@ -75,18 +75,20 @@ gitea:
     serviceMonitor:
       enabled: true      # kube-prometheus-stack discovers ServiceMonitors cluster-wide
 
-# ── Persistence (Longhorn RWX — git repo data shared across replicas) ────────
+# ── Persistence (local/emptyDir — single pod only; no replicas) ────────
+# Longhorn RWX volume was corrupted. Rebuild with local storage.
+# Once stable, can re-enable RWX and scale to replicas.
 persistence:
   enabled: true
-  storageClass: longhorn
+  storageClass: ""  # empty string = local/emptyDir, no storage backend
   size: 20Gi
   accessModes:
-    - ReadWriteMany
+    - ReadWriteOnce
 
 # ── Replicas ──────────────────────────────────────────────────────────────────
-# Postgres (CNPG) + Redis (session/cache/queue) + RWX storage means replicas
-# no longer need to serialize on a single pod like the old SQLite setup did.
-replicaCount: 2
+# Temporary: single pod with local storage while volume recovers.
+# Once stable, scale back to 2+ with RWX Longhorn storage.
+replicaCount: 1
 
 # ── Deployment strategy ────────────────────────────────────────────────────────
 deployment: