Quick Answer

# Run a command with nohup and send it to the background
nohup ./long-running-script.sh &

# Redirect output explicitly
nohup ./long-running-script.sh > output.log 2>&1 &

What You’re Trying to Do

You connected to a VPS over SSH, kicked off a long-running task, and it died the moment the connection dropped — this is exactly the problem nohup was built to solve.

When an SSH session ends, any process started in it normally receives a SIGHUP (hang up) signal and terminates along with it. nohup — short for “no hang up” — launches a command in a way that ignores SIGHUP, so the process keeps running even after you log out.

Environment

  • OS: Linux (verified on Ubuntu, CentOS, Debian, and other major distributions)
  • Shell: bash / zsh

Solution

1. Basic Syntax

nohup <command>

This alone ignores SIGHUP, but the command still runs in the foreground and blocks your terminal, so you’ll usually add & to send it to the background too.

nohup ./long-running-script.sh &

2. Redirecting Output

If you run nohup without &, stdout and stderr are automatically written to nohup.out in the current directory.

nohup ./long-running-script.sh &
cat nohup.out

To control where output goes, use explicit redirection.

# Combine stdout and stderr into one file
nohup ./long-running-script.sh > output.log 2>&1 &

# Split stdout and stderr
nohup ./long-running-script.sh > stdout.log 2> stderr.log &

# Discard all output
nohup ./long-running-script.sh > /dev/null 2>&1 &

3. Checking the Process ID

When run in the background, the process ID is printed immediately.

nohup ./long-running-script.sh &
[1] 12345

To check later, use jobs -l or ps.

jobs -l
ps aux | grep long-running-script

4. Killing the Process

kill 12345

Force-kill with -9 if needed.

kill -9 12345

5. Combining with disown

nohup ignores SIGHUP, but depending on your shell configuration, the job may still be tracked by the shell itself. To fully detach it from the shell, combine it with disown.

nohup ./long-running-script.sh &
disown

Now the process stays alive even if the shell that launched it exits, since it’s no longer in the shell’s job table.

6. Applying nohup to an Already-Running Process

You can’t retroactively attach nohup to a process that’s already running. In that case, use disown alone, or consider switching to a terminal multiplexer like screen or tmux.

# Detach a running job from the shell (not a full substitute for nohup)
disown -h %1

Common Errors

nohup: ignoring input and appending output to 'nohup.out'

This isn’t an error — it’s the normal message shown when you don’t redirect output explicitly. It just means stdin is ignored and output is being appended to nohup.out.

nohup: failed to run command: No such file or directory

Usually means the script or command wasn’t found, or it isn’t executable. Check the path and permissions.

ls -l ./long-running-script.sh
chmod +x ./long-running-script.sh

nohup.out Keeps Growing

If a process produces a lot of output without redirection, nohup.out can grow indefinitely. Redirect to /dev/null if you don’t need it, or set up log rotation with something like logrotate.

The Process Still Dies After SSH Disconnects

If you forgot the trailing &, the command stays in the foreground and still terminates when you close the SSH connection, nohup or not. Double-check that you actually backgrounded it.

FAQ

Q: What’s the difference between nohup and screen / tmux? nohup simply makes one command ignore SIGHUP. screen and tmux persist an entire session, letting you reattach and interact with it later. Use screen / tmux for long interactive work, and nohup for one-off batch jobs.

Q: Is nohup command & the same as command & disown? Similar, but not identical. nohup makes the process ignore the SIGHUP signal itself, while disown removes it from the shell’s job table. For maximum safety, combine both.

Q: How do I check whether a process started with nohup has finished? Run ps aux | grep <process-name> to see if it’s still running, or check the log file. If you need the exit code, have the script write it to the log itself on completion.

Q: Can I use nohup instead of a systemd service? It’s fine for temporary or ad hoc tasks. But if you need auto-start on boot, automatic restarts, and proper log management in production, defining a systemd service is the more reliable approach.

Q: Why does my process still die even though I used nohup? Common causes include forgetting the &, the whole process group being killed together, or the server itself being rebooted. Check the process tree with ps -ef --forest to investigate.

Recommended VPS / Hosting

Build your production environment on a reliable VPS: