diff --git a/apps/teslamate/manifests/cnpg-cluster.yaml b/apps/teslamate/manifests/cnpg-cluster.yaml
index 84e46a6..ed2b626 100644
--- a/apps/teslamate/manifests/cnpg-cluster.yaml
+++ b/apps/teslamate/manifests/cnpg-cluster.yaml
@@ -5,7 +5,7 @@ metadata:
   namespace: teslamate
 spec:
   imageName: ghcr.io/cloudnative-pg/postgresql:16
-  instances: 2
+  instances: 1
 
   # Resource limits to prevent OOM
   resources:
diff --git a/infrastructure/rook-ceph/cluster-values.yaml b/infrastructure/rook-ceph/cluster-values.yaml
index 3bfcdd0..5719563 100644
--- a/infrastructure/rook-ceph/cluster-values.yaml
+++ b/infrastructure/rook-ceph/cluster-values.yaml
@@ -120,6 +120,9 @@ cephClusterSpec:
         memory: "50Mi"
 
   # Storage configuration - explicit nodes with passthrough NVMe
+  # NOTE: talos-cp-02 excluded from nodes list because ceph-volume inventory
+  # hangs on its BIOS partition (lvs kernel I/O hang via nsenter on Talos).
+  # OSD 1 (nvme1n1) continues running from its existing deployment.
   storage:
     useAllNodes: false
     useAllDevices: false
@@ -127,9 +130,6 @@ cephClusterSpec:
       - name: "talos-cp-01"
         devices:
           - name: "nvme0n1"
-      - name: "talos-cp-02"
-        devices:
-          - name: "nvme1n1"
       - name: "talos-cp-03"
         devices:
           - name: "nvme0n1"