Ever started a job and thought “this is running on a bit longer than I expected”, then wondered whats going to happen when you go home and come back in to work tomorrow morning to find your remote session gone, which leaves you wondering “did that job complete?”.
Mmm, me too, which is where the nohup or disown commands come in. nohup (no hangup) is a well known job control utility which will prevent a process from reacting to the hangup signal when a shell is disconnected. Usually you’d preceed your actual command with it, e.g.
nohup rsync -auvc /Source/* /Destination/
but if your command is already running, you’re left wishing you’d nohup‘d it to start with – unless you’re running Solaris or AIX in which case the nohup command has a convenient -p switch to specify the process id (use ps -ef | grep rsync to obtain the PID of that long running data migration process, then nohup -p 9675 (or whatever the PID is of your running job).
If you’re not running Solaris or AIX, then pray you started the command in the bash shell (Linux default shell so more likely than not). If you did, then you can
to pause the current job running in the foreground, then use the
command to determine its job number (most likely 1 if there’s no other sysadmins running backgrounded jobs), then background the process with
to disconnect the process from the current shell. Running
again will show that your job is no longer in the list, but
will reveal that it is in fact still running.
Your shell can now be closed without the fear of your running job being killed with it. Yay.