Self-Hosting on a $5 Server: A Practical Weekend Guide

A few of my side projects need something a static site can't do: a scheduled job, a tiny database, a webhook endpoint. For years I'd reach for a free tier somewhere and then forget about it until it got shut down.

Now I run them on a single small virtual server that costs about the price of a fancy coffee each month. Here's how I set it up in a weekend, what I'd recommend, and what I got wrong.

What you'll need

• A small VPS from any reputable provider (1 vCPU, 1 GB RAM is plenty to start)
• A domain name you control
• A terminal and an SSH key
• About an afternoon

Step 1: Lock it down first

Before installing anything, secure the box. Bots will start knocking within minutes of it coming online — I'm not exaggerating.

# Create a non-root user and give it sudo
adduser stalin
usermod -aG sudo stalin
 
# Copy your SSH key to the new user
rsync --archive --chown=stalin:stalin ~/.ssh /home/stalin

Then edit /etc/ssh/sshd_config to disable password and root login:

PermitRootLogin no
PasswordAuthentication no

Restart SSH, and — this is important — test logging in from a second terminal before you close the first one. I locked myself out on my first attempt and had to use the provider's rescue console. Not my finest hour.

Step 2: A simple firewall

sudo ufw allow OpenSSH
sudo ufw allow 80,443/tcp
sudo ufw enable

Three commands, and everything except SSH and web traffic is closed.

Step 3: Automatic security updates

You will forget to update this server. Let it update itself:

sudo apt install unattended-upgrades
sudo dpkg-reconfigure --priority=low unattended-upgrades

Step 4: Containers and a reverse proxy

I run each project in its own container, with a reverse proxy in front that handles HTTPS certificates automatically. A minimal config looks like this:

uptime.stalink.com {
    reverse_proxy localhost:3001
}
 
links.stalink.com {
    reverse_proxy localhost:3002
}

Point a DNS record at the server's IP, and the proxy fetches a free certificate on first request. Adding a new project is now a two-line change.

Step 5: Backups (do this before you need them)

My rule: if it's not backed up off the server, it doesn't exist. A nightly cron job dumps databases and config to compressed archives and copies them to cheap object storage.

0 3 * * * /home/stalin/bin/backup.sh >> /var/log/backup.log 2>&1

Test a restore at least once. A backup you've never restored is just a hopeful file.

What runs on it today

• A tiny uptime monitor that pings my sites and messages me if one goes down
• My personal link archive
• A webhook that rebuilds this blog when I push a new post
• A small analytics script that counts page views without cookies or personal data

All of that, on one small server, using about 30% of its memory.

Mistakes I made

• Skipping the firewall "just for now." There is no "just for now" on the public internet.
• Running everything as root. Convenient right up until it isn't.
• Not monitoring disk space. Logs filled the disk in six weeks and took everything down. Log rotation is your friend.

Is it worth it?

For me, absolutely. I understand my tools better, I'm not dependent on a free tier's goodwill, and the cost is predictable and tiny. It's also genuinely fun — there's something satisfying about a small machine humming away somewhere, quietly doing exactly what you told it to.

If you've got a weekend and a side project that needs a home, give it a try. Just test that SSH login before you close the terminal.