Use pvesh API for cluster node discovery in install-sensor-proxy.sh

Replace brittle pvecm nodes CLI parsing with pvesh API calls. The old
approach used awk field positions ($4) which breaks across Proxmox
versions, locales, or output format changes.

Added get_cluster_node_names() helper that:
- Prefers pvesh get /cluster/status --output-format json (structured)
- Falls back to pvecm nodes CLI parsing if pvesh unavailable
- Uses python3 for JSON parsing (always available on Proxmox)

Related to #738

scripts/install-sensor-proxy.sh

@@ -785,6 +785,44 @@ REQUESTED_VERSION=""
 INSTALLER_CACHE_REASON=""
 DEFER_SOCKET_VERIFICATION=false
 
+# Get cluster node names via Proxmox API (pvesh) - returns one hostname per line
+# Falls back to pvecm CLI parsing if pvesh unavailable
+get_cluster_node_names() {
+    # Try pvesh API first (structured JSON, version-stable)
+    if command -v pvesh >/dev/null 2>&1; then
+        local json_output
+        if json_output=$(pvesh get /cluster/status --output-format json 2>/dev/null); then
+            # Parse JSON: extract "name" from entries where "type" is "node"
+            # Using python3 for reliable JSON parsing (always available on Proxmox)
+            if command -v python3 >/dev/null 2>&1; then
+                local names
+                names=$(echo "$json_output" | python3 -c '
+import sys, json
+try:
+    data = json.load(sys.stdin)
+    for item in data:
+        if item.get("type") == "node" and item.get("name"):
+            print(item["name"])
+except:
+    pass
+' 2>/dev/null)
+                if [[ -n "$names" ]]; then
+                    echo "$names"
+                    return 0
+                fi
+            fi
+        fi
+    fi
+
+    # Fallback: parse pvecm nodes CLI output (fragile, position-dependent)
+    if command -v pvecm >/dev/null 2>&1; then
+        pvecm nodes 2>/dev/null | awk '/^[[:space:]]+[0-9]/ && !/Qdevice/ {print $4}' || true
+        return 0
+    fi
+
+    return 1
+}
+
 cleanup_local_authorized_keys() {
     local auth_keys_file="/root/.ssh/authorized_keys"
     if [[ ! -f "$auth_keys_file" ]]; then
@@ -802,21 +840,19 @@ cleanup_local_authorized_keys() {
 
 cleanup_cluster_authorized_keys_manual() {
     local nodes=()
-    if command -v pvecm >/dev/null 2>&1; then
+    # Use get_cluster_node_names helper (prefers pvesh API, falls back to pvecm CLI)
-        # Prefer hostname resolution for multi-network clusters
     while IFS= read -r nodename; do
         [[ -z "$nodename" ]] && continue
         local resolved_ip
         resolved_ip=$(getent hosts "$nodename" 2>/dev/null | awk '{print $1; exit}')
         [[ -n "$resolved_ip" ]] && nodes+=("$resolved_ip")
-        done < <(pvecm nodes 2>/dev/null | awk '/^[[:space:]]+[0-9]/ && !/Qdevice/ {print $4}' || true)
+    done < <(get_cluster_node_names 2>/dev/null || true)
     # Fallback to pvecm status if hostname resolution didn't work
-        if [[ ${#nodes[@]} -eq 0 ]]; then
+    if [[ ${#nodes[@]} -eq 0 ]] && command -v pvecm >/dev/null 2>&1; then
         while IFS= read -r node_ip; do
             [[ -n "$node_ip" ]] && nodes+=("$node_ip")
         done < <(pvecm status 2>/dev/null | grep -vEi "qdevice" | awk '/0x[0-9a-f]+.*[0-9]+\.[0-9]+\.[0-9]+\.[0-9]+/ {for(i=1;i<=NF;i++) if($i ~ /^[0-9]+\.[0-9]+\.[0-9]+\.[0-9]+$/) print $i}' || true)
     fi
-    fi
 
     if [[ ${#nodes[@]} -eq 0 ]]; then
         cleanup_local_authorized_keys
@@ -2662,7 +2698,7 @@ if [[ -z "$HOST" ]]; then
     log_info "No specific host provided - cleaning up all cluster nodes"
 
     # Discover cluster nodes (prefer hostname resolution for multi-network clusters)
-    if command -v pvecm >/dev/null 2>&1; then
+    # Use get_cluster_node_names helper (prefers pvesh API, falls back to pvecm CLI)
     CLUSTER_NODES=""
     while IFS= read -r nodename; do
         [[ -z "$nodename" ]] && continue
@@ -2670,9 +2706,9 @@ if [[ -z "$HOST" ]]; then
         if [[ -n "$resolved_ip" ]]; then
             CLUSTER_NODES="${CLUSTER_NODES:+$CLUSTER_NODES }$resolved_ip"
         fi
-        done < <(pvecm nodes 2>/dev/null | awk '/^[[:space:]]+[0-9]/ && !/Qdevice/ {print $4}' || true)
+    done < <(get_cluster_node_names 2>/dev/null || true)
     # Fallback to pvecm status if hostname resolution didn't work
-        if [[ -z "$CLUSTER_NODES" ]]; then
+    if [[ -z "$CLUSTER_NODES" ]] && command -v pvecm >/dev/null 2>&1; then
         CLUSTER_NODES=$(pvecm status 2>/dev/null | grep -vEi "qdevice" | awk '/0x[0-9a-f]+.*[0-9]+\.[0-9]+\.[0-9]+\.[0-9]+/ {for(i=1;i<=NF;i++) if($i ~ /^[0-9]+\.[0-9]+\.[0-9]+\.[0-9]+$/) print $i}' || true)
     fi
 
@@ -2980,9 +3016,7 @@ fi
 PROXY_PUBLIC_KEY=$(cat "$PROXY_KEY_FILE")
 print_info "Proxy public key: ${PROXY_PUBLIC_KEY:0:50}..."
 
-# Discover cluster nodes
+# Discover cluster nodes using get_cluster_node_names helper (prefers pvesh API, falls back to pvecm CLI)
-if command -v pvecm >/dev/null 2>&1; then
-    # Extract node hostnames from pvecm nodes and resolve to IPs via /etc/hosts
 # This prefers management IPs over corosync ring IPs for multi-network clusters
 CLUSTER_NODES=""
 while IFS= read -r nodename; do
@@ -2993,15 +3027,17 @@ if command -v pvecm >/dev/null 2>&1; then
         CLUSTER_NODES="${CLUSTER_NODES:+$CLUSTER_NODES }$resolved_ip"
     else
         # Fallback: try extracting IP directly from pvecm status for this node
+        if command -v pvecm >/dev/null 2>&1; then
             status_ip=$(pvecm status 2>/dev/null | grep -E "0x[0-9a-f]+.*$nodename" | awk '{for(i=1;i<=NF;i++) if($i ~ /^[0-9]+\.[0-9]+\.[0-9]+\.[0-9]+$/) {print $i; exit}}')
             if [[ -n "$status_ip" ]]; then
                 CLUSTER_NODES="${CLUSTER_NODES:+$CLUSTER_NODES }$status_ip"
             fi
         fi
-    done < <(pvecm nodes 2>/dev/null | awk '/^[[:space:]]+[0-9]/ && !/Qdevice/ {print $4}' || true)
+    fi
+done < <(get_cluster_node_names 2>/dev/null || true)
 
-    # Fallback to original method if pvecm nodes didn't yield results
+# Fallback to pvecm status if get_cluster_node_names didn't yield results
-    if [[ -z "$CLUSTER_NODES" ]]; then
+if [[ -z "$CLUSTER_NODES" ]] && command -v pvecm >/dev/null 2>&1; then
     CLUSTER_NODES=$(pvecm status 2>/dev/null | grep -vEi "qdevice" | awk '/0x[0-9a-f]+.*[0-9]+\.[0-9]+\.[0-9]+\.[0-9]+/ {for(i=1;i<=NF;i++) if($i ~ /^[0-9]+\.[0-9]+\.[0-9]+\.[0-9]+$/) print $i}' || true)
 fi
 
@@ -3137,36 +3173,6 @@ if command -v pvecm >/dev/null 2>&1; then
         exit 1
     fi
 fi
-else
-    # Proxmox host but pvecm not available (shouldn't happen, but handle it)
-    print_warn "pvecm command not available"
-    print_info "Configuring SSH key for localhost..."
-
-    # Configure localhost as fallback
-    FORCED_CMD='command="/opt/pulse/sensor-proxy/bin/pulse-sensor-wrapper.sh",no-port-forwarding,no-X11-forwarding,no-agent-forwarding,no-pty'
-    AUTH_LINE="${FORCED_CMD} ${PROXY_PUBLIC_KEY} # pulse-managed-key"
-
-    configure_local_authorized_key "$AUTH_LINE"
-
-    # Add localhost to config file for allowlist validation
-    print_info "Updating proxy configuration for localhost fallback..."
-    LOCAL_IPS=$(hostname -I 2>/dev/null | tr ' ' '\n' | grep -v '^$' || echo "127.0.0.1")
-    # Collect all local IPs and localhost variants into array
-    all_nodes=()
-    for local_ip in $LOCAL_IPS; do
-        all_nodes+=("$local_ip")
-    done
-    # Always include localhost variants
-    all_nodes+=("127.0.0.1" "localhost")
-    if [[ ${#CONTROL_PLANE_ALLOWED_NODE_LIST[@]} -gt 0 ]]; then
-        all_nodes+=("${CONTROL_PLANE_ALLOWED_NODE_LIST[@]}")
-    fi
-    # Use helper function to safely update allowed_nodes (prevents duplicates on re-run)
-    if ! update_allowed_nodes "Localhost fallback configuration (pvecm unavailable)" "${all_nodes[@]}"; then
-        print_error "Failed to update allowed_nodes list"
-        exit 1
-    fi
-fi
 
 cleanup_inline_allowed_nodes