#!/usr/bin/env bash
# Check all Proxmox hosts: ping, SSH, health (disk, memory, load, services), VM/container counts.
# Usage: ./scripts/check-all-proxmox-hosts.sh
# Expects SSH key-based auth to root@<host>.

set -euo pipefail

SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
PROJECT_ROOT="$(cd "$SCRIPT_DIR/.." && pwd)"
# shellcheck source=/dev/null
source "${PROJECT_ROOT}/config/ip-addresses.conf" 2>/dev/null || true

PROXMOX_HOST_ML110="${PROXMOX_HOST_ML110:-192.168.11.10}"
PROXMOX_HOST_R630_01="${PROXMOX_HOST_R630_01:-192.168.11.11}"
PROXMOX_HOST_R630_02="${PROXMOX_HOST_R630_02:-192.168.11.12}"

SSH_OPTS="-o StrictHostKeyChecking=no -o ConnectTimeout=5 -o BatchMode=yes"
RED='\033[0;31m'
GREEN='\033[0;32m'
YELLOW='\033[1;33m'
CYAN='\033[0;36m'
NC='\033[0m'

log_section() { echo -e "\n${CYAN}━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━${NC}\n${CYAN}$1${NC}\n${CYAN}━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━${NC}\n"; }
ok() { echo -e "${GREEN}[✓]${NC} $1"; }
warn() { echo -e "${YELLOW}[⚠]${NC} $1"; }
fail() { echo -e "${RED}[✗]${NC} $1"; }

declare -a HOSTS=(
    "ml110:${PROXMOX_HOST_ML110}"
    "r630-01:${PROXMOX_HOST_R630_01}"
    "r630-02:${PROXMOX_HOST_R630_02}"
)

log_section "Proxmox hosts health check (all)"
echo "Started: $(date -Iseconds 2>/dev/null || date)"
echo ""

for entry in "${HOSTS[@]}"; do
    IFS=: read -r name ip <<< "$entry"
    echo -e "${CYAN}--- $name ($ip) ---${NC}"
    # Ping
    if ping -c 1 -W 2 "$ip" >/dev/null 2>&1; then
        ok "Ping: reachable"
    else
        fail "Ping: unreachable"
        echo ""
        continue
    fi
    # SSH + health + status
    out=$(ssh $SSH_OPTS root@"$ip" "
        echo \"PVEVER|$(pveversion -v 2>/dev/null | head -1 || echo 'unknown')\"
        echo \"HOSTNAME|$(hostname 2>/dev/null)\"
        echo \"UPTIME|$(uptime -p 2>/dev/null || uptime 2>/dev/null | sed 's/.*load average/load/')\"
        echo \"LOAD|\$(cat /proc/loadavg 2>/dev/null | cut -d' ' -f1-3)\"
        df_root=\$(df -h / 2>/dev/null | tail -1 | awk '{print \$5 \" used \" \$2 \" total\"}')
        echo \"DISK_ROOT|${df_root:-N/A}\"
        df_lv=\$(df -h /var/lib/vz 2>/dev/null | tail -1 | awk '{print \$5 \" used \" \$2 \" total\"}')
        [ -n \"\$df_lv\" ] && echo \"DISK_VZ|\$df_lv\" || true
        mem=\$(free -h 2>/dev/null | awk '/^Mem:/ {print \$3 \"/\" \$2 \" used\"}')
        echo \"MEMORY|\${mem:-N/A}\"
        pveproxy=\$(systemctl is-active pveproxy 2>/dev/null || echo 'inactive')
        pvedaemon=\$(systemctl is-active pvedaemon 2>/dev/null || echo 'inactive')
        pvestatd=\$(systemctl is-active pvestatd 2>/dev/null || echo 'inactive')
        echo \"PVEPROXY|\$pveproxy\"
        echo \"PVEDAEMON|\$pvedaemon\"
        echo \"PVESTATD|\$pvestatd\"
        port8006=\$(ss -tlnp 2>/dev/null | grep -c ':8006 ' || echo 0)
        echo \"PORT8006|\$port8006\"
        pct list 2>/dev/null | tail -n +2 | wc -l | xargs echo 'LXC_COUNT|'
        qm list 2>/dev/null | tail -n +2 | wc -l | xargs echo 'QEMU_COUNT|'
    " 2>/dev/null) || true
    if [ -n "$out" ]; then
        ok "SSH: OK"
        pve_ok=true
        while IFS= read -r line; do
            key="${line%%|*}"
            val="${line#*|}"
            case "$key" in
                PVEVER)    echo "  PVE: $val" ;;
                HOSTNAME)  echo "  Hostname: $val" ;;
                UPTIME)    echo "  Uptime: $val" ;;
                LOAD)      echo "  Load (1/5/15): $val" ;;
                DISK_ROOT) echo "  Disk (/) : $val" ;;
                DISK_VZ)   echo "  Disk (vz): $val" ;;
                MEMORY)    echo "  Memory: $val" ;;
                PVEPROXY)
                    if [ "$val" = "active" ]; then echo "  pveproxy: $val"; else echo -e "  pveproxy: ${RED}$val${NC}"; pve_ok=false; fi ;;
                PVEDAEMON)
                    if [ "$val" = "active" ]; then echo "  pvedaemon: $val"; else echo -e "  pvedaemon: ${RED}$val${NC}"; pve_ok=false; fi ;;
                PVESTATD)
                    if [ "$val" = "active" ]; then echo "  pvestatd: $val"; else echo -e "  pvestatd: ${RED}$val${NC}"; pve_ok=false; fi ;;
                PORT8006)
                    if [ "${val:-0}" -gt 0 ] 2>/dev/null; then echo "  Web UI (8006): listening"; else echo -e "  Web UI (8006): ${YELLOW}not listening${NC}"; fi ;;
                LXC_COUNT) echo "  LXC containers: $val" ;;
                QEMU_COUNT) echo "  QEMU VMs: $val" ;;
            esac
        done <<< "$out"
        # Overall health
        if [ "${pve_ok:-true}" = true ]; then
            ok "Health: OK"
        else
            warn "Health: one or more Proxmox services not active"
        fi
    else
        warn "SSH: failed or no key (BatchMode)"
    fi
    echo ""
done

log_section "Summary"
echo "Hosts: ml110 ($PROXMOX_HOST_ML110), r630-01 ($PROXMOX_HOST_R630_01), r630-02 ($PROXMOX_HOST_R630_02)"
echo "For full inventory: ./scripts/quick-proxmox-inventory.sh"
echo ""