Tag Archives: reboot

Networking, OS

NPM reboot workaround script

Leave a comment

I had an annoying “chicken and egg” problem with nginx proxy manager. It has hosts configurations that reference DNS names that are only reachable over VPN. It also hosts headscale, which is required for VPN to establish properly for said DNS entries. Brought up from scratch (say, after a reboot) NPM fails to launch because it can’t resolve the DNS entries. Headscale therefore isn’t reachable by the other nodes, and we’re stuck.

I developed a script to be run at boot to get around this annoying behavior of nginx proxy manager. It disables all other hosts in NPM except headscale, which doesn’t need DNS resolution as it resides on the same host. It then fires up NPM, and then tailscale, and waits to confirm the VPN is established and DNS resolution is working. Then it restores all the other hosts and restarts NPM. This script was 100% done myself, no AI assistance was used.

#!/bin/bash
# Script to properly bring things up after a reboot
# Name resolution doesn't happen until vpn VIP comes up
# NPM doesn't spin up until name resolution works
# vpn headscale is part of NPM
#
# Replace all VPN DNS host entries with 127.0.0.1
# Wait for NPM to come up
# Restore original host entries to get name resolution to work
# restart npm
#
# Run on a cron job @reboot. Ensure cron service is enabled.
# @reboot /docker/npm/reboot-script.sh | tee /tmp/reboot-script.log

# Variables
DOCKER_DIR=/docker/npm
CONF_DIR=$DOCKER_DIR/data/nginx/proxy_host
TEMP_DIR=/tmp/$(basename $CONF_DIR)
VPN_IP=100.1.1.1

# Log start of script
echo "Reboot script started on $(date)"

# Backup host entries
rsync -aP --delete $CONF_DIR $(dirname $TEMP_DIR) 

# Disable all hosts except for headscale
sed -i  's/\".*.<VPN_DNS_SUFFIX>\"/\"127.0.0.1\"/g' $CONF_DIR/*.conf
sed -i  's/http:\/\/.*.<VPN_DNS_SUFFIX>/http:\/\/127.0.0.1/g' $CONF_DIR/*.conf

# Listen on 127.0.0.1 for npm
sed -i "s/$VPN_IP/127.0.0.1/g" $DOCKER_DIR/docker-compose.yml

# Stop and restart npm
cd $CONF_DIR && docker compose down && docker compose up -d

# Wait for NPM to launch
until netstat -an|grep 0.0.0.0:80 >/dev/null; do echo "Waiting for NPM"; sleep 5; done; echo "NPM up"

# Restart tailscale
systemctl restart tailscaled

# Wait until VPN pings are successful
until ping -w 1 -c 1 <VPN_DNS_IP> >/dev/null; do echo "Waiting for VPN"; sleep 5; done; echo "VPN Successful"

# Restore config files
rsync -aP $TEMP_DIR/ $CONF_DIR
sed -i "s/127.0.0.1/$VPN_IP/g" $DOCKER_DIR/docker-compose.yml

# Restart npm
cd $CONF_DIR && docker compose restart 

echo "Reboot script completed on $(date)"

CLI, OS

Reboot a windows system remotely

Leave a comment

I recently came across the need to reboot my Windows 10 workstation remotely. I couldn’t RDP into the box  – it would stay stuck on the disclaimer splash screen. There was an OK button but it did nothing when clicked.

My solution was to initiate a reboot command from another Windows PC on the same network as my Windows 10 workstation. I figured out how to do so thanks to this article.

To remotely reboot yourself, log onto another Windows box and open up a command shell with an account that has administrative permissions on the machine you want to reboot. Then, issue this command:

shutdown -r -f -m \\<remote machine name> -t 30 -c "<message>"

The above command forces a reboot command (shutdown -r -f) on the remote machine (-m \\machine name), causes a countdown timer for the number of seconds specified (-t 30) and displays a message on the screen (-c “message”).

It solved my issue and was pretty easy to do. Handy.