Table of Contents

The nohup command runs a command immune to the hangup signal (SIGHUP).

It is useful when you need a process to continue running after:

  • Closing your terminal
  • Disconnecting SSH sessions
  • Ending an interactive shell

Overview

When a command is launched with nohup, standard behavior is:

  • Ignore SIGHUP
  • Redirect output to nohup.out if not explicitly redirected
  • Continue running in the background when combined with &

nohup is simple and portable for quick jobs, but does not provide full service management.

Syntax

nohup command [arguments...]

Common usage pattern:

nohup command [arguments...] > output.log 2>&1 &

Command Quick Reference

Command Purpose
nohup cmd & Run command detached from hangup signal
nohup cmd > out.log 2>&1 & Run in background with explicit logging
nohup bash -c "..." & Run a small command sequence detached
jobs -l Show background jobs in current shell
ps -fp <pid> Inspect launched process by PID

Core Workflows

Launch a Long-Running Command

nohup ./long-running-job.sh > job.log 2>&1 &

This is the most common pattern for SSH sessions and remote automation.

Capture the Process ID

nohup ./worker.sh > worker.log 2>&1 &
echo $!

$! is the PID of the most recent background command.

Run a Pipeline Through a Shell

nohup bash -c 'tar -czf backup.tgz /data && echo done' > backup.log 2>&1 &

Use bash -c when the command requires shell features such as &&, pipes, or redirects.

Monitor Process and Logs

ps -fp <pid>
tail -f job.log

Stop a nohup-Launched Process

kill -TERM <pid>

Use kill -KILL <pid> only if graceful termination fails.

Output and Redirection Behavior

If no redirection is provided:

  • Standard output is appended to nohup.out
  • Standard error is often redirected to standard output

Recommended explicit redirection:

nohup command > command.log 2>&1 &

This keeps logs predictable and avoids unexpected files in the working directory.

Useful Patterns

Run from Any Directory with Absolute Paths

nohup /usr/local/bin/sync-task --config /etc/sync/config.yml > /var/log/sync-task.log 2>&1 &

Start and Disown in Interactive Shells

nohup ./dev-server.sh > dev-server.log 2>&1 &
disown

disown removes the job from shell job tracking.

Use with Environment Variables

nohup env APP_ENV=prod ./app-worker > app-worker.log 2>&1 &

Examples

nohup ./long-running-job.sh &
nohup ./long-running-job.sh > job.log 2>&1 &
nohup bash -c 'python3 sync.py --once' > sync.log 2>&1 &

Troubleshooting

Command Stops Unexpectedly

nohup protects against SIGHUP, but not all failures. Check:

  1. Application-level errors in log files
  2. Resource limits (memory, disk, ulimit)
  3. Manual termination or system restart

nohup.out Is Growing Unexpectedly

You likely did not redirect output explicitly. Use:

nohup command > command.log 2>&1 &

Process Not Found After Disconnect

Verify command started successfully and capture PID immediately with $!. Also check logs for startup failures.

Permission Denied Writing Logs

Ensure target log path is writable by the current user, or run with proper privileges.

Best Practices

  1. Always redirect output to a named log file.
  2. Capture and store PID ($!) when launching background jobs.
  3. Prefer absolute paths in long-running detached commands.
  4. Use systemd (systemctl) for production services requiring restart policies and boot integration.
  5. Combine with log rotation for long-running jobs.

When to Use Alternatives

  • Use systemd services for production daemons.
  • Use tmux or screen when you want to keep an interactive session.
  • Use workload schedulers (cron, systemd timers, CI runners) for scheduled tasks.
  • ps: Inspect running process details
  • pgrep: Find matching process IDs
  • kill and pkill: Stop or signal background processes
  • systemctl: Manage long-running services via systemd
  • disown: Remove jobs from shell tracking
Edit this page