How to run your own program as a systemd service that starts at boot

For a program of yours (a Node.js app, a Python script, a bot) to start with the server and come back by itself if it dies, you write a unit file in /etc/systemd/system/, reload systemd and enable the service with systemctl enable --now. It is the native way on any modern Linux, with nothing extra to install. For Node.js apps there is also PM2 (keeping a Node.js app running with PM2); systemd covers everything else, and Node too.

Step by step

1 Run the program by hand first. If it does not start in the terminal, it will not start as a service. Note the program’s full path (which node or which python3) and the folder it runs from.
2 Create an unprivileged user for the service instead of running it as root: sudo useradd --system --home /opt/myapp --shell /usr/sbin/nologin myapp. If the program is compromised, the damage stays tied to that user.
3 Write the unit with sudo nano /etc/systemd/system/myapp.service[Unit]
Description=My application
After=network.target

[Service]
User=myapp
WorkingDirectory=/opt/myapp
ExecStart=/usr/bin/node /opt/myapp/server.js
Restart=on-failure
RestartSec=5
EnvironmentFile=/etc/myapp.env

[Install]
WantedBy=multi-user.target
4 Keep the secrets in a separate file, /etc/myapp.env, one NAME=value line per variable, and protect it: sudo chmod 600 /etc/myapp.env. See environment variables and secrets.
5 Tell systemd there is a new unit and start it: sudo systemctl daemon-reload then sudo systemctl enable --now myapp. “enable” makes it start with the server; “--now” starts it right away.
6 Confirm. systemctl status myapp should say “active (running)”, and journalctl -u myapp -f shows what the program writes, live.
7 Prove the reboot. Restart the server at a quiet hour and see whether the service returns. That is the only way to know it works.

The unit’s lines, in words

Line What it is for
After=network.target Starts only after the network is ready.
User= Who runs the program. Never root, unless it truly needs it.
ExecStart= The command, with the program’s full path. systemd uses neither your PATH nor your terminal.
Restart=on-failure Starts again if it ends with an error; does not if you stopped it on purpose.
EnvironmentFile= Reads variables from a file instead of writing them in the unit (which any user can read).
WantedBy=multi-user.target What makes enable start it at the server’s normal boot.
ExecStart is not a terminal. Redirections, && and ~ do not work as they do at the prompt; use absolute paths and, if you need several commands, put them in a script and run the script. If you see “status=203/EXEC”, the path is wrong: reading systemctl status.
Changed the unit and nothing changed? You are missing sudo systemctl daemon-reload. And a unit that gets “Permission denied” reading a folder is permissions or SELinux: file permissions.
Scheduling with a timer instead of cron: systemd schedules too (.timer files), but for simple jobs cron is still the most direct: when cron does not run.

Does the program run by hand and fail as a service, and you have read the log? Show us the error text and what changed since it worked.

Open a support ticket

SEE ALSO

A service will not start: reading systemctl status

Where the logs are on a Linux VPS

Keeping a Node.js app running with PM2

How far our support goes

RECOMMENDED PRODUCT

Web hosting with cPanel

Domain and SSL included, daily backups and the panel you already know. from £5.28/mo (3-year plan, with coupon)

See plans
  • 0 Users Found This Useful
Was this answer helpful?

Related Articles

First steps with your VPS A VPS is a server of your own: you choose the operating system and you have full root access. That also means the respon...
Connecting to a Windows VPS with Remote Desktop A Windows VPS is reached with Remote Desktop instead of SSH. The rest - how the server is delivered and what's your resp...
Common VPS and Cloud errors: SSH, memory, disk and reboots On a VPS the machine is yours: the system, the services and the firewall obey your orders - and your mistakes. These are...
Managed or unmanaged VPS: which to choose and who does what Two VPS with the same processor, the same memory and the same disk can be priced very differently. The difference is rar...
Reinstalling or changing your VPS operating system On shared hosting the operating system is not your choice - and just as well, because it is not your problem either. On...
Where are the logs on a Linux VPS: journalctl and /var/log On a Linux VPS the logs live in two places: in the systemd journal, read with journalctl, and in files under /var/log. O...