systemctl 命令完全指南

时间:2022-12-08 15:13:03

http://www.linuxidc.com/Linux/2015-07/120833.htm

Systemctl是一个systemd工具,主要负责控制systemd系统和服务管理器。

Systemd是一个系统管理守护进程、工具和库的集合,用于取代System V初始进程。Systemd的功能是用于集中管理和配置类UNIX系统。

在Linux生态系统中,Systemd被部署到了大多数的标准Linux发行版中,只有为数不多的几个发行版尚未部署。Systemd通常是所有其它守护进程的父进程,但并非总是如此。

systemctl 命令完全指南

使用Systemctl管理Linux服务

本文旨在阐明在运行systemd的系统上“如何控制系统和服务”。

Systemd初体验和Systemctl基础

1. 首先检查你的系统中是否安装有systemd并确定当前安装的版本

  1. # systemd --version
  2. systemd 215
  3. +PAM +AUDIT +SELINUX +IMA +SYSVINIT +LIBCRYPTSETUP +GCRYPT +ACL +XZ -SECCOMP -APPARMOR

上例中很清楚地表明,我们安装了215版本的systemd。

2. 检查systemd和systemctl的二进制文件和库文件的安装位置

  1. # whereis systemd
  2. systemd:/usr/lib/systemd /etc/systemd /usr/share/systemd /usr/share/man/man1/systemd.1.gz
  3. # whereis systemctl
  4. systemctl:/usr/bin/systemctl /usr/share/man/man1/systemctl.1.gz

3. 检查systemd是否运行

  1. # ps -eaf | grep [s]ystemd
  2. root 10016:27?00:00:00/usr/lib/systemd/systemd --switched-root --system --deserialize 23
  3. root 4441016:27?00:00:00/usr/lib/systemd/systemd-journald
  4. root 4691016:27?00:00:00/usr/lib/systemd/systemd-udevd
  5. root 5551016:27?00:00:00/usr/lib/systemd/systemd-logind
  6. dbus 5561016:27?00:00:00/bin/dbus-daemon --system --address=systemd:--nofork --nopidfile --systemd-activation

注意:systemd是作为父进程(PID=1)运行的。在上面带(-e)参数的ps命令输出中,选择所有进程,(-a)选择除会话前导外的所有进程,并使用(-f)参数输出完整格式列表(即 -eaf)。

也请注意上例中后随的方括号和例子中剩余部分。方括号表达式是grep的字符类表达式的一部分。

4. 分析systemd启动进程

  1. # systemd-analyze
  2. Startup finished in487ms(kernel)+2.776s(initrd)+20.229s(userspace)=23.493s

5. 分析启动时各个进程花费的时间

  1. # systemd-analyze blame
  2. 8.565s mariadb.service
  3. 7.991s webmin.service
  4. 6.095s postfix.service
  5. 4.311s httpd.service
  6. 3.926s firewalld.service
  7. 3.780s kdump.service
  8. 3.238s tuned.service
  9. 1.712s network.service
  10. 1.394s lvm2-monitor.service
  11. 1.126s systemd-logind.service
  12. ....

6. 分析启动时的关键链

  1. # systemd-analyze critical-chain
  2. The time after the unit is active or started is printed after the "@" character.
  3. The time the unit takes to start is printed after the "+" character.
  4. multi-user.target @20.222s
  5. └─mariadb.service @11.657s+8.565s
  6. └─network.target @11.168s
  7. └─network.service @9.456s+1.712s
  8. └─NetworkManager.service @8.858s+596ms
  9. └─firewalld.service @4.931s+3.926s
  10. └─basic.target @4.916s
  11. └─sockets.target @4.916s
  12. └─dbus.socket @4.916s
  13. └─sysinit.target @4.905s
  14. └─systemd-update-utmp.service @4.864s+39ms
  15. └─auditd.service @4.563s+301ms
  16. └─systemd-tmpfiles-setup.service @4.485s+69ms
  17. └─rhel-import-state.service @4.342s+142ms
  18. └─local-fs.target @4.324s
  19. └─boot.mount @4.286s+31ms
  20. └─systemd-fsck@dev-disk-by\x2duuid-79f594ad\x2da332\x2d4730\x2dbb5f\x2d85d19608096
  21. └─dev-disk-by\x2duuid-79f594ad\x2da332\x2d4730\x2dbb5f\x2d85d196080964.device@4

重要:Systemctl接受服务(.service),挂载点(.mount),套接口(.socket)和设备(.device)作为单元。

7. 列出所有可用单元

  1. # systemctl list-unit-files
  2. UNIT FILE STATE
  3. proc-sys-fs-binfmt_misc.automount static
  4. dev-hugepages.mount static
  5. dev-mqueue.mount static
  6. proc-sys-fs-binfmt_misc.mount static
  7. sys-fs-fuse-connections.mount static
  8. sys-kernel-config.mount static
  9. sys-kernel-debug.mount static
  10. tmp.mount disabled
  11. brandbot.path disabled
  12. .....

8. 列出所有运行中单元

  1. # systemctl list-units
  2. UNIT LOAD ACTIVE SUB DESCRIPTION
  3. proc-sys-fs-binfmt_misc.automount loaded active waiting ArbitraryExecutableFileFormatsFileSyste
  4. sys-devices-pc...0-1:0:0:0-block-sr0.device loaded active plugged VBOX_CD-ROM
  5. sys-devices-pc...:00:03.0-net-enp0s3.device loaded active plugged PRO/1000 MT DesktopAdapter
  6. sys-devices-pc...00:05.0-sound-card0.device loaded active plugged 82801AA AC'97 Audio Controller
  7. sys-devices-pc...:0:0-block-sda-sda1.device loaded active plugged VBOX_HARDDISK
  8. sys-devices-pc...:0:0-block-sda-sda2.device loaded active plugged LVM PV Qzyo3l-qYaL-uRUa-Cjuk-pljo-qKtX-VgBQ8
  9. sys-devices-pc...0-2:0:0:0-block-sda.device loaded active plugged VBOX_HARDDISK
  10. sys-devices-pl...erial8250-tty-ttyS0.device loaded active plugged /sys/devices/platform/serial8250/tty/ttyS0
  11. sys-devices-pl...erial8250-tty-ttyS1.device loaded active plugged /sys/devices/platform/serial8250/tty/ttyS1
  12. sys-devices-pl...erial8250-tty-ttyS2.device loaded active plugged /sys/devices/platform/serial8250/tty/ttyS2
  13. sys-devices-pl...erial8250-tty-ttyS3.device loaded active plugged /sys/devices/platform/serial8250/tty/ttyS3
  14. sys-devices-virtual-block-dm\x2d0.device loaded active plugged /sys/devices/virtual/block/dm-0
  15. sys-devices-virtual-block-dm\x2d1.device loaded active plugged /sys/devices/virtual/block/dm-1
  16. sys-module-configfs.device loaded active plugged /sys/module/configfs
  17. ...

9. 列出所有失败单元

  1. # systemctl --failed
  2. UNIT LOAD ACTIVE SUB DESCRIPTION
  3. kdump.service loaded failed failed Crash recovery kernel arming
  4. LOAD =Reflects whether the unit definition was properly loaded.
  5. ACTIVE =The high-level unit activation state, i.e. generalization of SUB.
  6. SUB =The low-level unit activation state, values depend on unit type.
  7. 1 loaded units listed.Pass--all to see loaded but inactive units, too.
  8. To show all installed unit files use'systemctl list-unit-files'.

10. 检查某个单元(如 cron.service)是否启用

  1. # systemctl is-enabled crond.service
  2. enabled

11. 检查某个单元或服务是否运行

  1. # systemctl status firewalld.service
  2. firewalld.service - firewalld -dynamic firewall daemon
  3. Loaded: loaded (/usr/lib/systemd/system/firewalld.service; enabled)
  4. Active: active (running) since Tue2015-04-2816:27:55 IST;34min ago
  5. Main PID:549(firewalld)
  6. CGroup:/system.slice/firewalld.service
  7. └─549/usr/bin/python -Es/usr/sbin/firewalld --nofork --nopid
  8. Apr2816:27:51 tecmint systemd[1]:Starting firewalld -dynamic firewall daemon...
  9. Apr2816:27:55 tecmint systemd[1]:Started firewalld -dynamic firewall daemon.

使用Systemctl控制并管理服务

12. 列出所有服务(包括启用的和禁用的)

  1. # systemctl list-unit-files --type=service
  2. UNIT FILE STATE
  3. arp-ethers.service disabled
  4. auditd.service enabled
  5. autovt@.service disabled
  6. blk-availability.service disabled
  7. brandbot.service static
  8. collectd.service disabled
  9. console-getty.service disabled
  10. console-shell.service disabled
  11. cpupower.service disabled
  12. crond.service enabled
  13. dbus-org.Fedoraproject.FirewallD1.service enabled
  14. ....

13. Linux中如何启动、重启、停止、重载服务以及检查服务(如 httpd.service)状态

  1. # systemctl start httpd.service
  2. # systemctl restart httpd.service
  3. # systemctl stop httpd.service
  4. # systemctl reload httpd.service
  5. # systemctl status httpd.service
  6. httpd.service -TheApache HTTP Server
  7. Loaded: loaded (/usr/lib/systemd/system/httpd.service; enabled)
  8. Active: active (running) since Tue2015-04-2817:21:30 IST;6s ago
  9. Process:2876ExecStop=/bin/kill -WINCH ${MAINPID}(code=exited, status=0/SUCCESS)
  10. Main PID:2881(httpd)
  11. Status:"Processing requests..."
  12. CGroup:/system.slice/httpd.service
  13. ├─2881/usr/sbin/httpd -DFOREGROUND
  14. ├─2884/usr/sbin/httpd -DFOREGROUND
  15. ├─2885/usr/sbin/httpd -DFOREGROUND
  16. ├─2886/usr/sbin/httpd -DFOREGROUND
  17. ├─2887/usr/sbin/httpd -DFOREGROUND
  18. └─2888/usr/sbin/httpd -DFOREGROUND
  19. Apr2817:21:30 tecmint systemd[1]:StartingTheApache HTTP Server...
  20. Apr2817:21:30 tecmint httpd[2881]: AH00558: httpd:Couldnot reliably determine the server's fully q...ssage
  21. Apr 28 17:21:30 tecmint systemd[1]: Started The Apache HTTP Server.
  22. Hint: Some lines were ellipsized, use -l to show in full.

注意:当我们使用systemctl的start,restart,stop和reload命令时,我们不会从终端获取到任何输出内容,只有status命令可以打印输出。

14. 如何激活服务并在启动时启用或禁用服务(即系统启动时自动启动服务)

  1. # systemctl is-active httpd.service
  2. # systemctl enable httpd.service
  3. # systemctl disable httpd.service

15. 如何屏蔽(让它不能启动)或显示服务(如 httpd.service)

  1. # systemctl mask httpd.service
  2. ln -s '/dev/null''/etc/systemd/system/httpd.service'
  3. # systemctl unmask httpd.service
  4. rm '/etc/systemd/system/httpd.service'

16. 使用systemctl命令杀死服务

  1. # systemctl kill httpd
  2. # systemctl status httpd
  3. httpd.service -TheApache HTTP Server
  4. Loaded: loaded (/usr/lib/systemd/system/httpd.service; enabled)
  5. Active: failed (Result:exit-code) since Tue2015-04-2818:01:42 IST;28min ago
  6. Main PID:2881(code=exited, status=0/SUCCESS)
  7. Status:"Total requests: 0; Current requests/sec: 0; Current traffic: 0 B/sec"
  8. Apr2817:37:29 tecmint systemd[1]: httpd.service:Got notification message from PID 2881, but recepti...bled.
  9. Apr2817:37:29 tecmint systemd[1]: httpd.service:Got notification message from PID 2881, but recepti...bled.
  10. Apr2817:37:39 tecmint systemd[1]: httpd.service:Got notification message from PID 2881, but recepti...bled.
  11. Apr2817:37:39 tecmint systemd[1]: httpd.service:Got notification message from PID 2881, but recepti...bled.
  12. Apr2817:37:49 tecmint systemd[1]: httpd.service:Got notification message from PID 2881, but recepti...bled.
  13. Apr2817:37:49 tecmint systemd[1]: httpd.service:Got notification message from PID 2881, but recepti...bled.
  14. Apr2817:37:59 tecmint systemd[1]: httpd.service:Got notification message from PID 2881, but recepti...bled.
  15. Apr2817:37:59 tecmint systemd[1]: httpd.service:Got notification message from PID 2881, but recepti...bled.
  16. Apr2818:01:42 tecmint systemd[1]: httpd.service: control process exited, code=exited status=226
  17. Apr2818:01:42 tecmint systemd[1]:Unit httpd.service entered failed state.
  18. Hint:Some lines were ellipsized,use-l to show in full.

使用Systemctl控制并管理挂载点

17. 列出所有系统挂载点

  1. # systemctl list-unit-files --type=mount
  2. UNIT FILE STATE
  3. dev-hugepages.mount static
  4. dev-mqueue.mount static
  5. proc-sys-fs-binfmt_misc.mount static
  6. sys-fs-fuse-connections.mount static
  7. sys-kernel-config.mount static
  8. sys-kernel-debug.mount static
  9. tmp.mount disabled

18. 挂载、卸载、重新挂载、重载系统挂载点并检查系统中挂载点状态

  1. # systemctl start tmp.mount
  2. # systemctl stop tmp.mount
  3. # systemctl restart tmp.mount
  4. # systemctl reload tmp.mount
  5. # systemctl status tmp.mount
  6. tmp.mount -TemporaryDirectory
  7. Loaded: loaded (/usr/lib/systemd/system/tmp.mount; disabled)
  8. Active: active (mounted) since Tue2015-04-2817:46:06 IST;2min48s ago
  9. Where:/tmp
  10. What: tmpfs
  11. Docs: man:hier(7)
  12. http://www.freedesktop.org/wiki/Software/systemd/APIFileSystems
  13. Process:3908ExecMount=/bin/mount tmpfs /tmp -t tmpfs -o mode=1777,strictatime (code=exited, status=0/SUCCESS)
  14. Apr2817:46:06 tecmint systemd[1]:MountingTemporaryDirectory...
  15. Apr2817:46:06 tecmint systemd[1]: tmp.mount:Directory/tmp to mount over isnot empty, mounting anyway.
  16. Apr2817:46:06 tecmint systemd[1]:MountedTemporaryDirectory.

19. 在启动时激活、启用或禁用挂载点(系统启动时自动挂载)

  1. # systemctl is-active tmp.mount
  2. # systemctl enable tmp.mount
  3. # systemctl disable tmp.mount

20. 在Linux中屏蔽(让它不能启用)或可见挂载点

  1. # systemctl mask tmp.mount
  2. ln -s '/dev/null''/etc/systemd/system/tmp.mount'
  3. # systemctl unmask tmp.mount
  4. rm '/etc/systemd/system/tmp.mount'

使用Systemctl控制并管理套接口

21. 列出所有可用系统套接口

  1. # systemctl list-unit-files --type=socket
  2. UNIT FILE STATE
  3. dbus.socket static
  4. dm-event.socket enabled
  5. lvm2-lvmetad.socket enabled
  6. rsyncd.socket disabled
  7. sshd.socket disabled
  8. syslog.socket static
  9. systemd-initctl.socket static
  10. systemd-journald.socket static
  11. systemd-shutdownd.socket static
  12. systemd-udevd-control.socket static
  13. systemd-udevd-kernel.socket static
  14. 11 unit files listed.

22. 在Linux中启动、重启、停止、重载套接口并检查其状态

  1. # systemctl start cups.socket
  2. # systemctl restart cups.socket
  3. # systemctl stop cups.socket
  4. # systemctl reload cups.socket
  5. # systemctl status cups.socket
  6. cups.socket - CUPS PrintingServiceSockets
  7. Loaded: loaded (/usr/lib/systemd/system/cups.socket; enabled)
  8. Active: active (listening) since Tue2015-04-2818:10:59 IST;8s ago
  9. Listen:/var/run/cups/cups.sock (Stream)
  10. Apr2818:10:59 tecmint systemd[1]:Starting CUPS PrintingServiceSockets.
  11. Apr2818:10:59 tecmint systemd[1]:Listening on CUPS PrintingServiceSockets.

23. 在启动时激活套接口,并启用或禁用它(系统启动时自启动)

  1. # systemctl is-active cups.socket
  2. # systemctl enable cups.socket
  3. # systemctl disable cups.socket

24. 屏蔽(使它不能启动)或显示套接口

  1. # systemctl mask cups.socket
  2. ln -s '/dev/null''/etc/systemd/system/cups.socket'
  3. # systemctl unmask cups.socket
  4. rm '/etc/systemd/system/cups.socket'

服务的CPU利用率(分配额)

25. 获取当前某个服务的CPU分配额(如httpd)

  1. # systemctl show -p CPUShares httpd.service
  2. CPUShares=1024

注意:各个服务的默认CPU分配份额=1024,你可以增加/减少某个进程的CPU分配份额。

26. 将某个服务(httpd.service)的CPU分配份额限制为2000 CPUShares/

  1. # systemctl set-property httpd.service CPUShares=2000
  2. # systemctl show -p CPUShares httpd.service
  3. CPUShares=2000

注意:当你为某个服务设置CPUShares,会自动创建一个以服务名命名的目录(如 httpd.service),里面包含了一个名为90-CPUShares.conf的文件,该文件含有CPUShare限制信息,你可以通过以下方式查看该文件:

  1. # vi /etc/systemd/system/httpd.service.d/90-CPUShares.conf
  2. [Service]
  3. CPUShares=2000

27. 检查某个服务的所有配置细节

  1. # systemctl show httpd
  2. Id=httpd.service
  3. Names=httpd.service
  4. Requires=basic.target
  5. Wants=system.slice
  6. WantedBy=multi-user.target
  7. Conflicts=shutdown.target
  8. Before=shutdown.target multi-user.target
  9. After=network.target remote-fs.target nss-lookup.target systemd-journald.socket basic.target system.slice
  10. Description=TheApache HTTP Server
  11. LoadState=loaded
  12. ActiveState=active
  13. SubState=running
  14. FragmentPath=/usr/lib/systemd/system/httpd.service
  15. ....

28. 分析某个服务(httpd)的关键链

  1. # systemd-analyze critical-chain httpd.service
  2. The time after the unit is active or started is printed after the "@" character.
  3. The time the unit takes to start is printed after the "+" character.
  4. httpd.service +142ms
  5. └─network.target @11.168s
  6. └─network.service @9.456s+1.712s
  7. └─NetworkManager.service @8.858s+596ms
  8. └─firewalld.service @4.931s+3.926s
  9. └─basic.target @4.916s
  10. └─sockets.target @4.916s
  11. └─dbus.socket @4.916s
  12. └─sysinit.target @4.905s
  13. └─systemd-update-utmp.service @4.864s+39ms
  14. └─auditd.service @4.563s+301ms
  15. └─systemd-tmpfiles-setup.service @4.485s+69ms
  16. └─rhel-import-state.service @4.342s+142ms
  17. └─local-fs.target @4.324s
  18. └─boot.mount @4.286s+31ms
  19. └─systemd-fsck@dev-disk-by\x2duuid-79f594ad\x2da332\x2d4730\x2dbb5f\x2d85d196080964.service@4.092s+149ms
  20. └─dev-disk-by\x2duuid-79f594ad\x2da332\x2d4730\x2dbb5f\x2d85d196080964.device@4.092s

29. 获取某个服务(httpd)的依赖性列表

  1. # systemctl list-dependencies httpd.service
  2. httpd.service
  3. ├─system.slice
  4. └─basic.target
  5. ├─firewalld.service
  6. ├─microcode.service
  7. ├─rhel-autorelabel-mark.service
  8. ├─rhel-autorelabel.service
  9. ├─rhel-configure.service
  10. ├─rhel-dmesg.service
  11. ├─rhel-loadmodules.service
  12. ├─paths.target
  13. ├─slices.target
  14. │├─-.slice
  15. │└─system.slice
  16. ├─sockets.target
  17. │├─dbus.socket
  18. ....

30. 按等级列出控制组

  1. # systemd-cgls
  2. ├─1/usr/lib/systemd/systemd --switched-root --system --deserialize 23
  3. ├─user.slice
  4. │└─user-0.slice
  5. │└─session-1.scope
  6. │├─2498 sshd: root@pts/0
  7. │├─2500-bash
  8. │├─4521 systemd-cgls
  9. │└─4522 systemd-cgls
  10. └─system.slice
  11. ├─httpd.service
  12. │├─4440/usr/sbin/httpd -DFOREGROUND
  13. │├─4442/usr/sbin/httpd -DFOREGROUND
  14. │├─4443/usr/sbin/httpd -DFOREGROUND
  15. │├─4444/usr/sbin/httpd -DFOREGROUND
  16. │├─4445/usr/sbin/httpd -DFOREGROUND
  17. │└─4446/usr/sbin/httpd -DFOREGROUND
  18. ├─polkit.service
  19. │└─721/usr/lib/polkit-1/polkitd --no-debug
  20. ....

31. 按CPU、内存、输入和输出列出控制组

  1. # systemd-cgtop
  2. PathTasks%CPU MemoryInput/s Output/s
  3. /831.0437.8M--
  4. /system.slice -0.1---
  5. /system.slice/mariadb.service 20.1---
  6. /system.slice/tuned.service 10.0---
  7. /system.slice/httpd.service 60.0---
  8. /system.slice/NetworkManager.service 1----
  9. /system.slice/atop.service 1----
  10. /system.slice/atopacct.service 1----
  11. /system.slice/auditd.service 1----
  12. /system.slice/crond.service 1----
  13. /system.slice/dbus.service 1----
  14. /system.slice/firewalld.service 1----
  15. /system.slice/lvm2-lvmetad.service 1----
  16. /system.slice/polkit.service 1----
  17. /system.slice/postfix.service 3----
  18. /system.slice/rsyslog.service 1----
  19. /system.slice/system-getty.slice/getty@tty1.service 1----
  20. /system.slice/systemd-journald.service 1----
  21. /system.slice/systemd-logind.service 1----
  22. /system.slice/systemd-udevd.service 1----
  23. /system.slice/webmin.service 1----
  24. /user.slice/user-0.slice/session-1.scope3----

控制系统运行等级

32. 启动系统救援模式

  1. # systemctl rescue
  2. Broadcast message from root@tecmint on pts/0(Wed2015-04-2911:31:18 IST):
  3. The system is going down to rescue mode NOW!

33. 进入紧急模式

  1. # systemctl emergency
  2. Welcome to emergency mode!After logging in, type "journalctl -xb" to view
  3. system logs,"systemctl reboot" to reboot,"systemctl default" to try again
  4. to boot intodefault mode.

34. 列出当前使用的运行等级

  1. # systemctl get-default
  2. multi-user.target

35. 启动运行等级5,即图形模式

  1. # systemctl isolate runlevel5.target
  2. # systemctl isolate graphical.target

36. 启动运行等级3,即多用户模式(命令行)

  1. # systemctl isolate runlevel3.target
  2. # systemctl isolate multiuser.target

36. 设置多用户模式或图形模式为默认运行等级

  1. # systemctl set-default runlevel3.target
  2. # systemctl set-default runlevel5.target

37. 重启、停止、挂起、休眠系统或使系统进入混合睡眠

  1. # systemctl reboot
  2. # systemctl halt
  3. # systemctl suspend
  4. # systemctl hibernate
  5. # systemctl hybrid-sleep

对于不知运行等级为何物的人,说明如下。

  • Runlevel 0 : 关闭系统
  • Runlevel 1 : 救援?维护模式
  • Runlevel 3 : 多用户,无图形系统
  • Runlevel 4 : 多用户,无图形系统
  • Runlevel 5 : 多用户,图形化系统
  • Runlevel 6 : 关闭并重启机器

到此为止吧。保持连线,进行评论。别忘了在下面的评论中为我们提供一些有价值的反馈哦。喜欢我们、与我们分享,求扩散。


via: http://www.tecmint.com/manage-services-using-systemd-and-systemctl-in-linux/

作者:Avishek Kumar 译者:GOLinux 校对:wxy

本文由 LCTT 原创翻译,Linux中国 荣誉推出

本文永久更新链接地址http://www.linuxidc.com/Linux/2015-07/120833.htm