Great article: http://www.cyberciti.biz/faq/unix-linux-disown-command-examples-usage-syntax/

Also this one:  http://unix.stackexchange.com/questions/3886/difference-between-nohup-disown-and

Can put a process into background using “command &“, typing “fg” will bring it to the foreground, you can press control-z to pause it and put it in the background, after that you probably want to unpause it while its still in the background (it will keep running in the back) type “bg“. Pressing control-C on a current process will cancel it completely. While a process is in the background, you can type “disown” to remove the job from the process list.

If you want to have the freedom of running commands in the background and being able to close the shell without killing the process you need to untie the shell (such as putty) from sending a HUP command to the process  when you close the shell. This is called “nohup”ing. You can untie it from shell (aka nohup) using “nohup command &“, all output of the commmand will go to file in the current working directory called “nohup.out”, you can redirect the output to a different file like so “nohup command 2> file2 > file1 &” or “nohup command 2>&1 >file1” (which is the same as “nohup command &> file1 &“. In the first redirection example, file1 gets normal output and file2 gets the error output, in the second and third example both normal & error output go to file1.

sidenote: untying a process from the shell means that when the shell closes it will not send it a hangup signal (hup), if it was still tied to the shell it would receive a hup message and it will close. The shell naturally throws hup signals when it itself is hupped(closed), it throws them at all processes that are running or stopped, but not to processes that were untied from the shell.

An alternative way to nohup a process is to start a process into the background and then disown it (with the -h): so first run this “command &“, that will start the commmmmand into the background but it will still be tied to the shell (so closing this putty session, for example, will kill the command you ran), now to untie it from the shell type “disown -h“.  This is useful if you launched a process into the background and decided later you want the process to continue to run when you close the shell.

Another alternative way to nohup a process is to first start the process like normal (into the foreground): “command“. Now put it into the background with control-z, it will be paused so type “bg” to start it up (it will still be in the background). Finally type “disown -h” to untie it from the shell. This is useful if you started a process and decided you need to close down the shell and still have the process run.

The last three paragraphs above talk about nohupping a process, its a great way to continue doing other things without having to worry about your background task ever closing, its a great safety blanket knowing that your process will run. For example your working on something super important remotely and storm rolls by where your at (not where the server is at), so you can nohup the process and if your computer (where your terminal shell is) gets shutoff you can have the assurance that your process still runs. There are other great tools that give you alot more features using this “nohupping” technique, I recommend checking out and fully learning the following three and in this order: “dtach” (yup thats detach without the e), “screen“, “tmux“. The big picture here is “Process Control” and I would have the following sections if I were to teach it in a logical order: “normal launching process”, control-c, control-z, “kill“, foreground and background using “bg“/”fg“, listing jobs with “jobs” and “jobs -l“, what it means to nohup and how to “nohup“, and then disowning/nohupping background task with “disown“, disowning/nohupping foreground tasks with “control-z,bg and disown -h“, “dtach” (thats detach without the e again), “screen” and “tmux“.

One thing I forgot to mention in the summmary, you can list the current jobs with “jobs” and if you want to see the process id type “jobs -l“. The rest of the article will be a more indepth look on all of this. Also it will show you how to deal with multiple processes. Everything above in the summary deals with one process in the background, if you have more then the following notations will become handy %n, where n is the job number such as %1, %2, which can be appended to most of the commands above to signify what job number you want to affect with “fg %2“,”bg %2“,”disown %2“, or “disown -h %2“.

NOTE: THE DIFFERENCE BETWEEN disown AND disown -h: you can use “disown” or “disown -h” to remove a process from the shell (so that when you turn off putty for example the process still runs – as side note to jog your head you can also use “nohup command &” to unie a process from the shell but thats not what is being asked here) – so whats the difference between “disown” and “disown -h“? The difference between them is that “disown” will untie the process from a shell (by supressing HUP/hang up messages) but it will also remove the process from the job list (seen with “jobs” or “jobs -l“), so you wont be able to control it with “bg” or “fg“, where as “disown -h” only unties the process from a shell (by supressing HUP/hang up messages), so that disown -h doesnt remove the process from the job list, so you will be able to use “bg” or “fg“. So “disown -h” and “disown” both untie a process from a shell, but “disown” without -h removes the process from the job list, where “disown -h” does not.

NOTE: WHY SHOULD I RUN “nohup command &” WITH THE BACKGROUND SYMBOL? Because “nohup command” redirects all output to a file, so might as well send the whole process to the background using “&“. Common practice is to do “nohup command &“, you never see “nohup command“

NOTE: WHATS THE DIFFERENCE BETWEEN disown AND disown -h AND nohup? Well “disown” and “disown -h” have in common that you have to start a process first and then get it into the background (or do it together “command &“) and then you can unattach it from the shell via “disown” or “disown -h” – nohup does all that in one move “nohup command &“. There are other differences too. nohup keeps processes in the job list (until you turn off the shell – when you come back to a shell the process still runs but your job list is gone).” disown -h” keeps processes in the jobs list and it also unties the process from a shell, however your output doesn’t get redirected to a file (it all stays on the terminal), with “disown” (and “disown -h“) your stderr and stdout go to the terminal (so it can get confusing if you have a couple of apps/processes disowned with “disown” or “disown -h” as they all will be writing to the terminal/screen at the same time). With “disown” & “disown -h” & “nohup” If you close the shell and return, your process will still be running but your job list will be blank just like I mentioned happens with nohup previously. The between regular “disown” (so not “disown -h“) and “nohup” is that regular “disown” removes the job from the job list upon entering the command “disown” where as “nohup” & “disown -h” does not, and the other difference “nohup” redirects output to a file (nohup.out) where as “disown” (and “disown -h“) does not, but the similarity is that all three untie the process from a shell.

– disown: untie from shell (so can turn off shell & process still runs); untie from stdin (so keyboard is not listened to by the command); stdout and stderr go to terminal; removed from job list;

– disown -h: untie from shell (so can turn off shell & process still runs); untie from stdin (so keyboard is not listened to by the command); stdout and stderr go to terminal; (side note: notice its not removed from the job list, so it can be listed with “jobs” and “jobs -l”, up until you get out of the shell, upon return to a new shell you will not see the process with “jobs” and “jobs -l”)

NOTE: WHAT DOES UNTYING A PROCESS FROM THE SHELL MEAN? When you close a shell, it sends a HUP/hangup signal to all of its processes (stopped and running processes) so they all close. When you untie a process from the shell, that process will not get that HUP/hangup signal when the shell closes, thus allowing you to continue working. A shell is launched when you login with something like xterm,ssh or putty, so when you close out of lets say putty, all processes will get shut that you launched (unless they were untied from the shell using nohup,disown, or any of the other tools that I lightly mentioned: dtach,screen,tmux)

NOTE: WHAT DOES UNTYING A PROCESS FROM THE TERMINAL MEAN? It just means basically that the output will not go to the terminal / your screen. So normally stderr and stdout are the terminal your working on (so that you see messages on the screen) and stdin is the keyboard so that it accepts your keystrokes. When you fully untie a process from the terminal, that process will no longer obey commands from your keyboard and it will no longer throw stuff on the screen (from stderr or from stdin). nohup unties processes from the terminal, all output from stderr and stdin goes to a file called nohup.out, also all input from the keyboard is ignored, its essentially impossible to send it keyboard commands (so this brings up an interesting question ill mention in a second). When you disown a process it doesn’t fully untie from the terminal, its stdin from the keyboard is untied, but its output from stderr and stdout still goes to the screen. Notice how disown/disown -h/nohup all untie the process from the terminal in some way (disown/disown -h only untie it from stdin/keybaord, and nohup fully unties a process so that not only stdin/keyboard is supressed, but stdout & stderr go to a file), but they are only tools to untie a process from a shell. So how do you untie a process from a shell, but not untie it from the terminal? The answer is with shell sessions such as those made with dtach,screen, or tmux – they can make a shell session which is untied from the shell, but you still have full shell control, when you close the putty, your session continues to run (and so do its processes), when you come back to the shell (your processes are still running) and normally (with using disown/disown -h/nohup) you wouldnt be able to control those processes (tie a terminal back to it so that keyboard gets stdin, and terminal/your display gets stdout & stderr), however with dtach/screen/tmux when you come back to a shell, you can log back into your session. Dtach keeps sessions in sizeless files called sockets files, and its as simple as it gets. Screen and tmux do so without socket files (hell maybe they exist somewhere in /tmp or /var – i didnt look), but they allow for sessions to be created in panes/panels or in different tabs (called windows). So with screen & tmux you can literally have 10 bash shells running under one session (its pretty cool), and all of those 10 bash shells are untied from putty (so when you close putty, your 10 bash shells and whatever processes they are running dont close). Check out articles on dtach/screen/tmux on google and infotinks.

---

---

---

---

                LOG BACK INTO THE SHELL: and check the size of the file a few times – if its increasing then the disown worked and the command is still running

NOTE2: Example – Running SOMECOMMAND at step0, that doesnt output to the screen. This is easier to work with because you can see the disown and exit commands

---

---

1. dtach – this starts a bash/sh session in another process so if you exit out of your shell it still runs

       Quick lesson on dtach

        To start a new sesssion type the following (note it makes a socket file called dtach.session):  dtach -A /tmp/dtach.session /bin/bash

         CONTROL+\ (key above enter key) = this gets you out of your dtach session and back in the shell you originally were in

         You can now close out of your shell (putty, or xterm, etc.) and it will still run

         To get back in your session dtach -A /tmp/dtach.session

         To end the dtach session, either restart or get back in the session and then type cd / and then exit to close the session once and for all then remove the socket file

        NOTE1: It doesnt matter if you call it dtach.session, i could of just called it sally without the .dtach. the /bin/bash argument is needed but you can put another shell like

2. screen – this is like dtach but with more “dtachs” you can even split the screen horizontaly and verticaly

            Google it to find out how to use it – just google “how to install screen linux” and “screen linux cheatsheet”

3. byobu – this is like screen but with a cool looking bash screen border and everything and easier to use keys

            Google it to find out how to use it – just google “how to install byobu” and “how to use byobu”