Linux 技巧:让进程在后台可靠运行的几种方法【转】

咱们常常会碰到这样的问题,用 telnet/ssh 登陆了远程的 Linux 服务器,运行了一些耗时较长的任务, 结果却因为网络的不稳定致使任务中途失败。如何让命令提交后不受本地关闭终端窗口/网络断开链接的干扰呢?下面举了一些例子, 您能够针对不一样的场景选择不一样的方式来处理这个问题。html

nohup/setsid/&

场景:

若是只是临时有一个命令须要长时间运行,什么方法能最简便的保证它在后台稳定运行呢?linux

解决方法:

咱们知道,当用户注销(logout)或者网络断开时,终端会收到 HUP(hangup)信号从而关闭其全部子进程。所以,咱们的解决办法就有两种途径:要么让进程忽略 HUP 信号,要么让进程运行在新的会话里从而成为不属于此终端的子进程。windows

1. nohupbash

nohup 无疑是咱们首先想到的办法。顾名思义,nohup 的用途就是让提交的命令忽略 hangup 信号。让咱们先来看一下 nohup 的帮助信息:服务器

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
NOHUP(1)                        User Commands                        NOHUP(1)
 
NAME
        nohup - run a command immune to hangups, with output to a non-tty
 
SYNOPSIS
        nohup COMMAND [ARG]...
        nohup OPTION
 
DESCRIPTION
        Run COMMAND, ignoring hangup signals.
 
        --help display this help and exit
 
        --version
               output version information and exit

可见,nohup 的使用是十分方便的,只需在要处理的命令前加上 nohup 便可,标准输出和标准错误缺省会被重定向到 nohup.out 文件中。通常咱们可在结尾加上"&"来将命令同时放入后台运行,也可用">filename 2>&1"来更改缺省的重定向文件名。网络

nohup 示例
1
2
3
4
5
6
7
[root@pvcent107 ~]# nohup ping www.ibm.com &
[1] 3059
nohup: appending output to `nohup.out'
[root@pvcent107 ~]# ps -ef |grep 3059
root      3059   984  0 21:06 pts/3    00:00:00 ping www.ibm.com
root      3067   984  0 21:06 pts/3    00:00:00 grep 3059
[root@pvcent107 ~]#

2。setsidsession

nohup 无疑能经过忽略 HUP 信号来使咱们的进程避免中途被中断,但若是咱们换个角度思考,若是咱们的进程不属于接受 HUP 信号的终端的子进程,那么天然也就不会受到 HUP 信号的影响了。setsid 就能帮助咱们作到这一点。让咱们先来看一下 setsid 的帮助信息:app

1
2
3
4
5
6
7
8
9
10
SETSID(8)                 Linux Programmer’s Manual                 SETSID(8)
 
NAME
        setsid - run a program in a new session
 
SYNOPSIS
        setsid program [ arg ... ]
 
DESCRIPTION
        setsid runs a program in a new session.

可见 setsid 的使用也是很是方便的,也只需在要处理的命令前加上 setsid 便可。less

setsid 示例
1
2
3
4
5
[root@pvcent107 ~]# setsid ping www.ibm.com
[root@pvcent107 ~]# ps -ef |grep www.ibm.com
root     31094     1  0 07:28 ?        00:00:00 ping www.ibm.com
root     31102 29217  0 07:29 pts/4    00:00:00 grep www.ibm.com
[root@pvcent107 ~]#

值得注意的是,上例中咱们的进程 ID(PID)为31094,而它的父 ID(PPID)为1(即为 init 进程 ID),并非当前终端的进程 ID。请将此例与nohup 例中的父 ID 作比较。

3。&

这里还有一个关于 subshell 的小技巧。咱们知道,将一个或多个命名包含在“()”中就能让这些命令在子 shell 中运行中,从而扩展出不少有趣的功能,咱们如今要讨论的就是其中之一。

当咱们将"&"也放入“()”内以后,咱们就会发现所提交的做业并不在做业列表中,也就是说,是没法经过jobs来查看的。让咱们来看看为何这样就能躲过 HUP 信号的影响吧。

subshell 示例
1
2
3
4
5
[root@pvcent107 ~]# (ping www.ibm.com &)
[root@pvcent107 ~]# ps -ef |grep www.ibm.com
root     16270     1  0 14:13 pts/4    00:00:00 ping www.ibm.com
root     16278 15362  0 14:13 pts/4    00:00:00 grep www.ibm.com
[root@pvcent107 ~]#

从上例中能够看出,新提交的进程的父 ID(PPID)为1(init 进程的 PID),并非当前终端的进程 ID。所以并不属于当前终端的子进程,从而也就不会受到当前终端的 HUP 信号的影响了。

disown

场景:

咱们已经知道,若是事先在命令前加上 nohup 或者 setsid 就能够避免 HUP 信号的影响。可是若是咱们未加任何处理就已经提交了命令,该如何补救才能让它避免 HUP 信号的影响呢?

解决方法:

这时想加 nohup 或者 setsid 已经为时已晚,只能经过做业调度和 disown 来解决这个问题了。让咱们来看一下 disown 的帮助信息:

1
2
3
4
5
6
7
8
9
10
11
disown [-ar] [-h] [jobspec ...]
     Without options, each jobspec is  removed  from  the  table  of
     active  jobs.   If  the -h option is given, each jobspec is not
     removed from the table, but is marked so  that  SIGHUP  is  not
     sent  to the job if the shell receives a SIGHUP.  If no jobspec
     is present, and neither the -a nor the -r option  is  supplied,
     the  current  job  is  used.  If no jobspec is supplied, the -a
     option means to remove or mark all jobs; the -r option  without
     a  jobspec  argument  restricts operation to running jobs.  The
     return value is 0 unless a jobspec does  not  specify  a  valid
     job.

能够看出,咱们能够用以下方式来达成咱们的目的。

  • disown -h jobspec来使某个做业忽略HUP信号。
  • disown -ah 来使全部的做业都忽略HUP信号。
  • disown -rh 来使正在运行的做业忽略HUP信号。

须要注意的是,当使用过 disown 以后,会将把目标做业从做业列表中移除,咱们将不能再使用jobs来查看它,可是依然可以用ps -ef查找到它。

可是还有一个问题,这种方法的操做对象是做业,若是咱们在运行命令时在结尾加了"&"来使它成为一个做业并在后台运行,那么就万事大吉了,咱们能够经过jobs命令来获得全部做业的列表。可是若是并无把当前命令做为做业来运行,如何才能获得它的做业号呢?答案就是用 CTRL-z(按住Ctrl键的同时按住z键)了!

CTRL-z 的用途就是将当前进程挂起(Suspend),而后咱们就能够用jobs命令来查询它的做业号,再用bg jobspec来将它放入后台并继续运行。须要注意的是,若是挂起会影响当前进程的运行结果,请慎用此方法。

disown 示例1(若是提交命令时已经用“&”将命令放入后台运行,则能够直接使用“disown”)
1
2
3
4
5
6
7
8
9
[root@pvcent107 build]# cp -r testLargeFile largeFile &
[1] 4825
[root@pvcent107 build]# jobs
[1]+  Running                 cp -i -r testLargeFile largeFile &
[root@pvcent107 build]# disown -h %1
[root@pvcent107 build]# ps -ef |grep largeFile
root      4825   968  1 09:46 pts/4    00:00:00 cp -i -r testLargeFile largeFile
root      4853   968  0 09:46 pts/4    00:00:00 grep largeFile
[root@pvcent107 build]# logout
disown 示例2(若是提交命令时未使用“&”将命令放入后台运行,可以使用 CTRL-z 和“bg”将其放入后台,再使用“disown”)
1
2
3
4
5
6
7
8
9
10
11
12
[root@pvcent107 build]# cp -r testLargeFile largeFile2
 
[1]+  Stopped                 cp -i -r testLargeFile largeFile2
[root@pvcent107 build]# bg %1
[1]+ cp -i -r testLargeFile largeFile2 &
[root@pvcent107 build]# jobs
[1]+  Running                 cp -i -r testLargeFile largeFile2 &
[root@pvcent107 build]# disown -h %1
[root@pvcent107 build]# ps -ef |grep largeFile2
root      5790  5577  1 10:04 pts/3    00:00:00 cp -i -r testLargeFile largeFile2
root      5824  5577  0 10:05 pts/3    00:00:00 grep largeFile2
[root@pvcent107 build]#

screen

场景:

咱们已经知道了如何让进程免受 HUP 信号的影响,可是若是有大量这种命令须要在稳定的后台里运行,如何避免对每条命令都作这样的操做呢?

解决方法:

此时最方便的方法就是 screen 了。简单的说,screen 提供了 ANSI/VT100 的终端模拟器,使它可以在一个真实终端下运行多个全屏的伪终端。screen 的参数不少,具备很强大的功能,咱们在此仅介绍其经常使用功能以及简要分析一下为何使用 screen 可以避免 HUP 信号的影响。咱们先看一下 screen 的帮助信息:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
SCREEN(1)                                                           SCREEN(1)
 
NAME
        screen - screen manager with VT100/ANSI terminal emulation
 
SYNOPSIS
        screen [ -options ] [ cmd [ args ] ]
        screen -r [[pid.]tty[.host]]
        screen -r sessionowner/[[pid.]tty[.host]]
 
DESCRIPTION
        Screen  is  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 functions from the  ISO  6429  (ECMA
        48,  ANSI  X3.64)  and ISO 2022 standards (e.g. insert/delete line and
        support for multiple character sets).  There is a  scrollback  history
        buffer  for  each virtual terminal and a copy-and-paste mechanism that
        allows moving text regions between windows.

使用 screen 很方便,有如下几个经常使用选项:

  • screen -dmS session name来创建一个处于断开模式下的会话(并指定其会话名)。
  • screen -list 来列出全部会话。
  • screen -r session name来从新链接指定会话。
  • 用快捷键CTRL-a d 来暂时断开当前会话。
screen 示例
1
2
3
4
5
6
7
[root@pvcent107 ~]# screen -dmS Urumchi
[root@pvcent107 ~]# screen -list
There is a screen on:
         12842.Urumchi   (Detached)
1 Socket in /tmp/screens/S-root.
 
[root@pvcent107 ~]# screen -r Urumchi

当咱们用“-r”链接到 screen 会话后,咱们就能够在这个伪终端里面随心所欲,不再用担忧 HUP 信号会对咱们的进程形成影响,也不用给每一个命令前都加上“nohup”或者“setsid”了。这是为何呢?让我来看一下下面两个例子吧。

1. 未使用 screen 时新进程的进程树
1
2
3
4
5
6
7
8
9
[root@pvcent107 ~]# ping www.google.com &
[1] 9499
[root@pvcent107 ~]# pstree -H 9499
init─┬─Xvnc
      ├─acpid
      ├─atd
      ├─2*[sendmail]
      ├─sshd─┬─sshd───bash───pstree
      │       └─sshd───bash───ping

咱们能够看出,未使用 screen 时咱们所处的 bash 是 sshd 的子进程,当 ssh 断开链接时,HUP 信号天然会影响到它下面的全部子进程(包括咱们新创建的 ping 进程)。

2. 使用了 screen 后新进程的进程树
1
2
3
4
5
6
7
8
9
[root@pvcent107 ~]# screen -r Urumchi
[root@pvcent107 ~]# ping www.ibm.com &
[1] 9488
[root@pvcent107 ~]# pstree -H 9488
init─┬─Xvnc
      ├─acpid
      ├─atd
      ├─screen───bash───ping
      ├─2*[sendmail]

而使用了 screen 后就不一样了,此时 bash 是 screen 的子进程,而 screen 是 init(PID为1)的子进程。那么当 ssh 断开链接时,HUP 信号天然不会影响到 screen 下面的子进程了。

总结

如今几种方法已经介绍完毕,咱们能够根据不一样的场景来选择不一样的方案。nohup/setsid 无疑是临时须要时最方便的方法,disown 能帮助咱们来过后补救当前已经在运行了的做业,而 screen 则是在大批量操做时不二的选择了。

 

转自

Linux 技巧:让进程在后台可靠运行的几种方法

https://www.ibm.com/developerworks/cn/linux/l-cn-nohup/

相关文章
相关标签/搜索