mirror of
https://github.com/community-scripts/ProxmoxVE.git
synced 2026-10-07 03:21:00 +02:00
Jason O'Brien
aaa341b6b2
The CT check resolved the IP by running ip inside the guest via pct exec, which depends on the guest having it installed. If not, this always comes back empty, so the container is restarted every cycle despite answering ping. lxc-info reports the addresses from the host without running anything inside the guest. Fixes #17685
196 lines
6.6 KiB
Bash
196 lines
6.6 KiB
Bash
#!/usr/bin/env bash
|
|
|
|
# Copyright (c) 2021-2026 tteck
|
|
# Author: tteck (tteckster)
|
|
# License: MIT
|
|
# https://github.com/community-scripts/ProxmoxVE/raw/main/LICENSE
|
|
|
|
clear
|
|
cat <<"EOF"
|
|
__ ___ _ __ ___ ____
|
|
/ |/ /___ ____ (_) /_____ _____ / | / / /
|
|
/ /|_/ / __ \/ __ \/ / __/ __ \/ ___/ / /| | / / /
|
|
/ / / / /_/ / / / / / /_/ /_/ / / / ___ |/ / /
|
|
/_/ /_/\____/_/ /_/_/\__/\____/_/ /_/ |_/_/_/
|
|
|
|
EOF
|
|
|
|
# Telemetry
|
|
source <(curl -fsSL "${COMMUNITY_SCRIPTS_CORE_URL:-https://raw.githubusercontent.com/community-scripts/core/main}/api/api.func") 2>/dev/null || true
|
|
declare -f init_tool_telemetry &>/dev/null && init_tool_telemetry "monitor-all" "pve"
|
|
|
|
add() {
|
|
echo -e "\n IMPORTANT: Tag-Based Monitoring Enabled"
|
|
echo "Only VMs and containers with the tag 'mon-restart' will be automatically restarted by this service."
|
|
echo
|
|
echo "🔧 How to add the tag:"
|
|
echo " → Proxmox Web UI: Go to VM/CT → Options → Tags → Add 'mon-restart'"
|
|
echo " → CLI: qm set <vmid> -tags mon-restart"
|
|
echo " pct set <ctid> -tags mon-restart"
|
|
echo
|
|
|
|
while true; do
|
|
read -p "This script will add Monitor All to Proxmox VE. Proceed (y/n)? " yn
|
|
case $yn in
|
|
[Yy]*) break ;;
|
|
[Nn]*) exit ;;
|
|
*) echo "Please answer yes or no." ;;
|
|
esac
|
|
done
|
|
|
|
cat <<'EOF' >/usr/local/bin/ping-instances.sh
|
|
#!/usr/bin/env bash
|
|
|
|
# Read excluded instances from command line arguments
|
|
excluded_instances=("$@")
|
|
echo "Excluded instances: ${excluded_instances[@]}"
|
|
|
|
value_of() { sed -n "s/^$1:[[:space:]]*//p" <<<"$config" | head -n1; }
|
|
|
|
while true; do
|
|
|
|
for instance in $(pct list | awk 'NR>1 {print $1}'; qm list | awk 'NR>1 {print $1}'); do
|
|
# Skip excluded instances
|
|
if [[ " ${excluded_instances[@]} " =~ " ${instance} " ]]; then
|
|
echo "Skipping $instance because it is excluded"
|
|
continue
|
|
fi
|
|
|
|
# Determine type and read the current config directly from Proxmox's pmxcfs.
|
|
# Ignore snapshot sections, matching the current config shown by pct/qm config.
|
|
if [ -r "/etc/pve/lxc/$instance.conf" ]; then
|
|
type="ct"
|
|
config_file="/etc/pve/lxc/$instance.conf"
|
|
elif [ -r "/etc/pve/qemu-server/$instance.conf" ]; then
|
|
type="vm"
|
|
config_file="/etc/pve/qemu-server/$instance.conf"
|
|
else
|
|
echo "Skipping $instance because its config is no longer readable"
|
|
continue
|
|
fi
|
|
config=$(sed '/^\[/,$d' "$config_file")
|
|
|
|
# Both are booleans that Proxmox also writes as 0, so the value decides and
|
|
# not the presence of the key.
|
|
[ "$(value_of onboot)" = "1" ] && onboot="true" || onboot="false"
|
|
[ "$(value_of template)" = "1" ] && template="true" || template="false"
|
|
|
|
if [ "$onboot" != "true" ]; then
|
|
echo "Skipping $instance because it is set not to boot"
|
|
continue
|
|
elif [ "$template" == "true" ]; then
|
|
echo "Skipping $instance because it is a template"
|
|
continue
|
|
fi
|
|
|
|
# Tags are semicolon separated, so match a whole entry instead of a
|
|
# substring of a longer tag such as mon-restart-disabled.
|
|
tags=";$(value_of tags | tr -d '[:space:]');"
|
|
if [[ "$tags" != *";mon-restart;"* ]]; then
|
|
echo "Skipping $instance because it does not have 'mon-restart' tag"
|
|
continue
|
|
fi
|
|
|
|
# Responsiveness check and restart if needed
|
|
if [ "$type" == "vm" ]; then
|
|
if ! qm status "$instance" 2>/dev/null | grep -q "status: running"; then
|
|
echo "$(date): VM $instance is not running, starting..."
|
|
qm start "$instance" >/dev/null 2>&1
|
|
elif qm guest cmd "$instance" ping >/dev/null 2>&1; then
|
|
echo "VM $instance is responsive via guest agent"
|
|
else
|
|
echo "$(date): VM $instance is not responding to agent ping, restarting..."
|
|
qm stop "$instance" >/dev/null 2>&1
|
|
sleep 5
|
|
qm start "$instance" >/dev/null 2>&1
|
|
fi
|
|
else
|
|
if ! pct status "$instance" 2>/dev/null | grep -q "status: running"; then
|
|
echo "$(date): CT $instance is not running, starting..."
|
|
pct start "$instance" >/dev/null 2>&1
|
|
continue
|
|
fi
|
|
# Read the IP from the host side; pct exec depends on tools inside the
|
|
# guest, which minimal/OCI images may lack.
|
|
IP=$(lxc-info -n "$instance" -iH 2>/dev/null | awk '/^[0-9.]+$/ && !/^(127|169\.254)\./ {print; exit}')
|
|
if [ -z "$IP" ] || ! ping -c 1 -W 2 "$IP" >/dev/null 2>&1; then
|
|
echo "$(date): CT $instance is not responding, restarting..."
|
|
pct stop "$instance" >/dev/null 2>&1
|
|
sleep 5
|
|
pct start "$instance" >/dev/null 2>&1
|
|
else
|
|
echo "CT $instance is responsive"
|
|
fi
|
|
fi
|
|
done
|
|
|
|
echo "$(date): Pausing for 5 minutes..."
|
|
sleep 300
|
|
|
|
done >/var/log/ping-instances.log 2>&1
|
|
EOF
|
|
|
|
touch /var/log/ping-instances.log
|
|
chmod +x /usr/local/bin/ping-instances.sh
|
|
|
|
# The service loops with its own five minute sleep, so the timer that earlier
|
|
# versions installed could only ever start a unit that was already running.
|
|
if [[ -f /etc/systemd/system/ping-instances.timer ]]; then
|
|
systemctl disable -q --now ping-instances.timer 2>/dev/null || true
|
|
rm -f /etc/systemd/system/ping-instances.timer
|
|
fi
|
|
|
|
cat <<EOF >/etc/systemd/system/ping-instances.service
|
|
[Unit]
|
|
Description=Ping instances every 5 minutes and restart if necessary
|
|
After=pve-cluster.service
|
|
Wants=pve-cluster.service
|
|
|
|
[Service]
|
|
Type=simple
|
|
# To exclude specific instances, pass IDs to ExecStart, e.g.:
|
|
# ExecStart=/usr/local/bin/ping-instances.sh 100 200
|
|
# Instances must also have the 'mon-restart' tag to be monitored
|
|
|
|
ExecStart=/usr/local/bin/ping-instances.sh
|
|
Restart=always
|
|
StandardOutput=file:/var/log/ping-instances.log
|
|
StandardError=file:/var/log/ping-instances.log
|
|
|
|
[Install]
|
|
WantedBy=multi-user.target
|
|
EOF
|
|
|
|
systemctl daemon-reload
|
|
systemctl enable -q --now ping-instances.service
|
|
clear
|
|
echo -e "\n Monitor All installed."
|
|
echo "📄 To view logs: cat /var/log/ping-instances.log"
|
|
echo "⚙️ Make sure your VMs or containers have the 'mon-restart' tag to be monitored."
|
|
}
|
|
|
|
remove() {
|
|
systemctl disable -q --now ping-instances.timer 2>/dev/null || true
|
|
systemctl disable -q --now ping-instances.service
|
|
rm -f /etc/systemd/system/ping-instances.service
|
|
rm -f /etc/systemd/system/ping-instances.timer
|
|
rm -f /usr/local/bin/ping-instances.sh
|
|
rm -f /var/log/ping-instances.log
|
|
echo "Monitor All removed from Proxmox VE"
|
|
}
|
|
|
|
OPTIONS=(Add "Add Monitor-All to Proxmox VE"
|
|
Remove "Remove Monitor-All from Proxmox VE")
|
|
|
|
CHOICE=$(whiptail --backtitle "Proxmox VE Helper Scripts" --title "Monitor-All for Proxmox VE" --menu "Select an option:" 10 58 2 \
|
|
"${OPTIONS[@]}" 3>&1 1>&2 2>&3)
|
|
|
|
case $CHOICE in
|
|
"Add") add ;;
|
|
"Remove") remove ;;
|
|
*)
|
|
echo "Exiting..."
|
|
exit 0
|
|
;;
|
|
esac
|