Transferred from: http://www.ibm.com/developerworks/cn/linux/l-cn-nohup/
Want the process to remain running after disconnecting? If the process is already running, how can it be remedied? If there are a large number of such requirements how to simplify the operation?
We often encounter such problems, log on to the remote Linux server with TELNET/SSH, run some long-time tasks, the result of the network instability caused by the task Midway failure. How do i make the command submit without being disturbed by the local shutdown of the terminal window/network disconnection? Here are some examples where you can choose different ways to handle this problem for different scenarios.
nohup/setsid/& Scene:
What is the easiest way to make sure it runs stably in the background if only a temporary command takes a long time to run?
Hangup the name of the
In earlier versions of Unix, each terminal was communicated via modem and system. When the user logout, the modem hangs up on the phone. Similarly, when the modem disconnects, it sends a hangup signal to the terminal to notify it to close all child processes.
Workaround:
We know that when the user logs off (logout) or the network disconnects, the terminal receives a HUP (hangup) signal to close all its child processes. Therefore, our solution has two ways: either let the process ignore the HUP signal, or let the process run in a new session to become a child process that does not belong to this terminal.
1. Nohup
Nohup is undoubtedly the way we think first. As the name implies, Nohup's purpose is to let the submitted command ignore the hangup signal. Let's take a look at Nohup's help information:
NOHUP (1) User Commands NOHUP (1) NAME nohup-run a command immune to hangups, with output to a non-ttysynopsis< C3/>nohup COMMAND [ARG] ... Nohup optiondescription Run COMMAND, ignoring hangup signals. --help Display this Help and exit --version output version information and exit
It can be seen that the use of nohup is very convenient, just add nohup before the command to be processed, standard output and standard errors are redirected to the Nohup.out file. In general, we can add "&" at the end to run the command in the background as well as ">filename 2>&1"
to change the default redirect file name.
nohup Example
[[email protected] ~]# nohup ping www.ibm.com &[1] 3059nohup:appending output to ' nohup.out ' [[email protected] ~]# PS -ef |grep 3059root 3059 984 0 21:06 pts/3 00:00:00 ping www.ibm.comroot 3067 984 0 21:06 pts/3 00:00:00 grep 3059[[email protected] ~]#
2. Setsid
Nohup can undoubtedly make our process avoid interruption by ignoring the HUP signal, but if we think in a different way, if our process is not part of the sub-process of the terminal receiving the HUP signal, then naturally it will not be affected by the HUP signal. Setsid can help us do that. Let's take a look at Setsid's help information:
Setsid (8) Linux Programmer ' s Manual setsid (8) NAME Setsid-run A program in a new sessionsynopsis Setsid program [ARG ...] DESCRIPTION Setsid runs a program in a new session.
Visible Setsid is also very convenient to use, but also only need to deal with the command before adding Setsid can.
Setsid Example
[[email protected] ~]# setsid ping www.ibm.com[[email protected] ~]# ps-ef |grep www.ibm.comroot 31094 1 > 0 07:28? 00:00:00 Ping www.ibm.comroot 31102 29217 0 07:29 pts/4 00:00:00 grep www.ibm.com[[email protected] ~]#
It is worth noting that our process ID (PID) in the example above is 31094, and its parent ID (PPID) is 1 (that is, the Init process ID), not the process ID of the current terminal. Compare this example to the parent ID in the nohup example.
3. &
Here's a little tip about Subshell. We know that the inclusion of one or more names in the "()" allows these commands to run in a child shell, which expands a lot of interesting functionality, and one of the things we're going to talk about right now.
When we put "&" into "()", we will find that the submitted job is not in the job list, that is, it cannot be jobs
viewed. Let's see why it's possible to avoid the effects of HUP signals.
Subshell Example
[[Email protected] ~]# (ping www.ibm.com &) [[email protected] ~]# ps-ef |grep www.ibm.comroot 16270 1 4> 0 14:13 pts/4 00:00:00 ping www.ibm.comroot 16278 15362 0 14:13 pts/4 00:00:00 grep www.ibm.com[ [Email protected] ~]#
As can be seen from the example above, the parent ID (PPID) of the newly submitted process is 1 (PID of the Init process) and is not the process ID of the current terminal. Therefore, it does not belong to the sub-process of the current terminal, so it will not be affected by the HUP signal of the current terminal.
Back to top of page
Disown scene:
We already know that the effects of HUP signals can be avoided by adding nohup or setsid prior to the command. But if we have already submitted an order without any processing, how can we remedy it to avoid the effects of the HUP signal?
Workaround:
At this time to add nohup or setsid is too late, only through job scheduling and disown to solve the problem. Let's take a look at disown's help information:
Disown [-ar] [-h] [jobspec ...] Without options, each jobspec was removed from the table ofactive jobs. If the-h option is a given, each jobspec was notremoved from the table, but was marked so that SIGHUP is
notsent to the job if the shell receives a SIGHUP. If no Jobspecis present, and neither the-a nor the-r option is supplied,the current job is
used. If no Jobspec is supplied, the-aoption means to remove or mark all jobs; The-r option Withouta jobspec argument restricts operation to running jobs. Thereturn value is 0 unless a jobspec does not specify a validjob.
As can be seen, we can achieve our goal in the following way.
Flexible use of ctrl-z
In our daily work, we can use the ctrl-z to suspend the current process to a background pause, perform some other action, and then use FG to back up the suspended process to the foreground (or BG to put the pending process in the background) to continue running. This allows us to flexibly switch between running multiple tasks within one terminal, which is especially useful when debugging code. Since the code editor is suspended to the background and then re-placed back, the cursor position remains at the location of the last suspension, avoiding the hassle of repositioning.
- Use
disown -h jobspec
to make a job ignore the HUP signal.
- Use
disown -ah
to make all jobs ignore the hup signal.
- Use
disown -rh
to make the running job ignore the hup signal.
It is important to note that when disown is used, the target job will be removed from the job list and we will no longer be jobs
able to use it to view it, but we can still ps -ef
find it.
However, there is a problem, the operation of this method is the job, if we run the command at the end of the "&" to make it a job and run in the background, then it is all right, we can jobs
get a list of all jobs by command. But if the current command is not running as a job, how can I get its job number? The answer is to use CTRL-Z (hold down the CTRL key while holding down the Z key)!
The purpose of Ctrl-z is to suspend the current process (Suspend), then we can use the jobs
command to query its job number, then use bg jobspec
it to put it in the background and continue to run. It is important to note that this method is used with caution if the suspension affects the running results of the current process.
Disown Example 1 (You can use "disown" directly if you have already used "&" to run the command in the background when you submit the command)
[[email protected] build]# cp-r testlargefile largefile &[1] 4825[[email protected] build]# jobs[1]+ running
cp-i-R testlargefile largefile &[[email protected] build]# disown-h%1[[email protected] build]# ps-ef |grep larg Efileroot 4825 968 1 09:46 pts/4 00:00:00 cp-i-R testlargefile largefileroot 4853 968 0 09:46 PTS/4 00:00:00 grep largefile[[email protected] build]# Logout
Disown Example 2 (if the command was submitted without using "&" to run it in the background, you can use Ctrl-z and "BG" to put it in the background and use "disown")
[Email protected] build]# cp-r testlargefile largefile2[1]+ Stopped cp-i-R testlargefile Largefile2[[email PR Otected] build]# BG%1[1]+ cp-i-R testlargefile largeFile2 &[[email protected] build]# jobs[1]+ Running CP- I-r testlargefile largeFile2 &[[email protected] build]# disown-h%1[[email protected] build]# ps-ef |grep largefil E2root 5790 5577 1 10:04 pts/3 00:00:00 cp-i-R testlargefile largefile2root 5824 5577 0 10:05 pts/3 00:00:00 grep largefile2[[email protected] build]#
Back to top of page
Screen scene:
We already know how to keep the process free of HUP signals, but if there are a lot of such commands that need to be run in a stable background, how do you avoid doing this for every command?
Workaround:
The most convenient way to do this is screen. Simply put, screen provides the ansi/vt100 terminal emulator, which enables it to run multiple full screens of pseudo-terminals under a real terminal. Screen has a lot of parameters and is very powerful, so we'll just describe its common functions and briefly analyze why using screen can avoid the effects of HUP signals. Let's take a look at screen's help information:
Screen (1) screens (1) NAME Screen-screen Manager with Vt100/ansi terminal emulationsynopsis screen [- Options] [cmd [args]] screen-r [[PID.] Tty[.host]] screen-r sessionowner/[[pid. Tty[.host]]description screen was a full-screen window manager that multiplexes a physical Terminal between several processes (typically interactive shells). Each virtual terminal provides the functions of a DEC VT100 terminal and , in addition, several control func tions from the ISO 6429 (ECMA , ANSI X3.64) and ISO 2022 standards (e.g. insert/ Delete line and support for multiple character sets). There is a scrollback the history of buffer for each virtual terminal and a copy-and-paste mechanism That allows moving text regions between windows.
Using screen is convenient and has several common options:
- Use
screen -dmS session name
to create a session in disconnected mode (and specify its session name).
- Use
screen -list
to list all sessions.
- Use
screen -r session name
to reconnect to the specified session.
- Use the shortcut key
CTRL-a d
to temporarily disconnect the current session.
Screen Example
[[email protected] ~]# SCREEN-DMS urumchi[[email protected] ~]# Screen-listthere is a screens on: 12842.Urumchi ( Detached) 1 Socket in/tmp/screens/s-root. [Email protected] ~]# screen-r Urumchi
When we connect to the screen session with "-R", we can do anything in this pseudo-terminal, no longer worry about the HUP signal will affect our process, and do not have to add "nohup" or "setsid" before each command. What is this for? Let me take a look at the following two examples.
1. Process tree without using the screen's newly process
[[email protected] ~]# ping www.google.com &[1] 9499[[email protected] ~]# pstree-h 9499init─┬─xvnc ├─acpid ├─ATD ├─2*[sendmail] ├─sshd─┬─sshd───bash───pstree │ └─sshd───bash───ping
As we can see, the bash we're in when we're not using screen is an sshd subprocess, and when SSH disconnects, the HUP signal naturally affects all of the sub-processes underneath it (including our newly created ping process).
2. Process tree using the new process after screen
[[email protected] ~]# screen-r urumchi[[email protected] ~]# ping www.ibm.com &[1] 9488[[email protected] ~]# Pstree -H 9488init─┬─xvnc ├─acpid ├─atd ├─screen───bash───ping ├─2*[sendmail]
Instead of using screen, bash is the child process of screen, and screen is the child of Init (PID 1). Then when SSH disconnects, the HUP signal naturally does not affect the sub-processes under screen.
Back to top of page
Summarize
Now that several methods have been introduced, we can choose different scenarios according to different scenarios. Nohup/setsid is undoubtedly the most convenient method for temporary needs, disown can help us to remedy the job that is currently running, and screen is the choice of a large batch operation.
Linux tips: Several ways to keep processes running reliably in the background