fix(ha): split health.sh VIP check into LAN (NFS) and storage (iSCSI)

Storage VIP (192.168.20.229) is on vmbr2 — an internal-only bridge with no
physical uplink, unreachable from the LAN. Test it via SSH to the active node
(which has VLAN 20 connectivity) instead of directly. LAN VIP (192.168.2.229)
continues to be tested directly from wherever the script runs.

Also adds VIP_LAN variable to config section.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01J8djTWdXVzXZc99iujU6T2

scripts/ha/health.sh

@@ -11,7 +11,8 @@ NODE1="${NODE1:-ha-server-1}"
 NODE2="${NODE2:-ha-server-2}"
 NODE1_IP="${NODE1_IP:-192.168.2.228}"   # vars.haServer1Ip
 NODE2_IP="${NODE2_IP:-192.168.2.227}"   # vars.haServer2Ip
-VIP="${VIP:-192.168.20.229}"            # vars.haServerVip
+VIP="${VIP:-192.168.20.229}"            # vars.haServerVip (storage-client, VLAN 20 — internal only)
+VIP_LAN="${VIP_LAN:-192.168.2.229}"    # vars.haServerLanVip (LAN, VLAN 2 — reachable from workstation)
 XFS_MOUNT="${XFS_MOUNT:-/srv/ha-data}"  # vars.haStorageRoot
 HA_USER="${HA_USER:-nixos}"
 # ──────────────────────────────────────────────────────────────────────────
@@ -149,25 +150,45 @@ check_mount() {
 $REACHABLE_1 && check_mount "$NODE1" n1 || echo "  $NODE1  [OFFLINE]"
 $REACHABLE_2 && check_mount "$NODE2" n2 || echo "  $NODE2  [OFFLINE]"
 
-# ── Service ports via VIP ──────────────────────────────────────────────────
+# ── LAN VIP (NFS) — reachable from workstation ────────────────────────────────
-section "Services via VIP ($VIP)"
+section "LAN VIP ($VIP_LAN) — NFS"
-
+if ping -c1 -W2 "$VIP_LAN" >/dev/null 2>&1; then
-check_port() {
-  local name=$1 port=$2
-  if bash -c "echo >/dev/tcp/${VIP}/${port}" 2>/dev/null; then
-    printf "  %-10s  port %-5s  OK\n" "$name" "$port"
-  else
-    printf "  %-10s  port %-5s  UNREACHABLE\n" "$name" "$port"
-  fi
-}
-
-if ping -c1 -W2 "$VIP" >/dev/null 2>&1; then
   echo "  Ping       OK"
 else
   echo "  Ping       UNREACHABLE"
 fi
-check_port "NFS"    2049
+if bash -c "echo >/dev/tcp/${VIP_LAN}/2049" 2>/dev/null; then
-check_port "iSCSI"  3260
+  printf "  %-10s  port %-5s  OK\n" "NFS" "2049"
+else
+  printf "  %-10s  port %-5s  UNREACHABLE\n" "NFS" "2049"
+fi
+
+# ── Storage VIP (iSCSI) — VLAN 20 internal bridge, tested via active node ─────
+section "Storage VIP ($VIP) — iSCSI (via ${ACTIVE_NODE:-unknown})"
+
+run_active_raw() {
+  local active_ip=""
+  [[ "$ACTIVE_NODE" == "$NODE1" ]] && active_ip="$NODE1_IP"
+  [[ "$ACTIVE_NODE" == "$NODE2" ]] && active_ip="$NODE2_IP"
+  [[ -z "$active_ip" ]] && return 1
+  ssh -i ~/.ssh/id_ed25519 -o StrictHostKeyChecking=no -o ConnectTimeout=5 \
+    "${HA_USER}@${active_ip}" "$@" 2>/dev/null
+}
+
+if [[ -z "$ACTIVE_NODE" ]]; then
+  echo "  Cannot determine active node — skipping"
+else
+  if run_active_raw "ping -c1 -W2 '$VIP' >/dev/null 2>&1"; then
+    echo "  Ping       OK"
+  else
+    echo "  Ping       UNREACHABLE"
+  fi
+  if run_active_raw "bash -c 'echo >/dev/tcp/${VIP}/3260' 2>/dev/null"; then
+    printf "  %-10s  port %-5s  OK\n" "iSCSI" "3260"
+  else
+    printf "  %-10s  port %-5s  UNREACHABLE\n" "iSCSI" "3260"
+  fi
+fi
 
 echo ""
 echo "════════════════════════════════════════════════════"