Gracefully shutdown an entire Proxmox VE cluster. Stops HA services (pve-ha-lrm, pve-ha-crm) on all nodes first, then shuts down remote nodes before the local node. Supports --dry-run. Requires SSH key exchange between nodes (default in Proxmox clusters) and node hostnames resolvable via /etc/hosts or DNS.

cluster-shutdown-all.sh

#!/bin/bash
# cluster-shutdown-all - Éteint proprement tout le cluster Proxmox
# Stoppe le HA sur tous les nœuds puis les éteint un par un
# Requiert les clés SSH root échangées entre nœuds (par défaut dans un cluster Proxmox)
# Usage: cluster-shutdown-all [--dry-run]

set -e

DRY_RUN=false
[ "$1" = "--dry-run" ] && DRY_RUN=true

LOCAL=$(hostname)
NODES=$(ls -1 /etc/pve/nodes/)

# Vérifier que le nœud local est bien dans la liste
if ! echo "$NODES" | grep -q "^${LOCAL}$"; then
    echo "ERROR: Local node $LOCAL not found in cluster nodes"
    exit 1
fi

echo "=== cluster-shutdown-all ==="
echo "Local node: $LOCAL"
echo "Detected nodes:"
for node in $NODES; do
    if [ "$node" = "$LOCAL" ]; then
        echo "  - $node (local)"
    else
        echo "  - $node (remote)"
    fi
done

if $DRY_RUN; then
    for node in $NODES; do
        if [ "$node" != "$LOCAL" ]; then
            echo "[DRY-RUN] Would stop pve-ha-lrm pve-ha-crm on $node via SSH"
        fi
    done
    echo "[DRY-RUN] Would stop pve-ha-lrm pve-ha-crm on $LOCAL"
    for node in $NODES; do
        if [ "$node" != "$LOCAL" ]; then
            echo "[DRY-RUN] Would shutdown $node via SSH"
        fi
    done
    echo "[DRY-RUN] Would shutdown $LOCAL (last)"
    echo "=== Dry run complete, no action taken ==="
    exit 0
fi

# Confirmation
echo ""
read -p "Are you sure you want to shutdown the entire cluster? (yes/no): " CONFIRM
if [ "$CONFIRM" != "yes" ]; then
    echo "Aborted."
    exit 0
fi

# Stopper le HA sur les nœuds distants
for node in $NODES; do
    if [ "$node" != "$LOCAL" ]; then
        echo "Stopping HA on $node..."
        ssh "$node" "systemctl stop pve-ha-lrm pve-ha-crm" &
    fi
done

echo "Stopping HA locally..."
systemctl stop pve-ha-lrm pve-ha-crm

# Attendre que les stops SSH finissent
wait

# Éteindre les nœuds distants
for node in $NODES; do
    if [ "$node" != "$LOCAL" ]; then
        echo "Shutting down $node..."
        ssh "$node" "/sbin/shutdown -h now" &
    fi
done

# Attendre que les commandes SSH soient envoyées
sleep 2

# Éteindre le nœud local en dernier
echo "Shutting down $LOCAL..."
/sbin/shutdown -h now