使用 systemd-analyze 命令可以查看 Linux 系统在启动过程中每个服务的耗时情况, 方便我们排查是哪个环节导致系统启动缓慢, 以下是整理的常用命令参数和效果. 例子中一下子就可以定位到是 gssproxy.service 服务启动耗时过长.

systemd-analyze blame Print list of running units ordered by time to init

$ systemd-analyze blame29.912s gssproxy.service11.280s update-motd.service4.250s network.service426ms cloud-init-local.service302ms cloud-init.service276ms dracut-initqueue.service247ms cloud-config.service206ms cloud-final.service201ms systemd-udev-settle.service183ms initrd-switch-root.service175ms postfix.service158ms lvm2-monitor.service

systemd-analyze critical-chain Print a tree of the time critical chain of units

$ systemd-analyze critical-chainThe time after the unit is active or started is printed after the "@" character.
The time the unit takes to start is printed after the "+" character.graphical.target @30.350s
└─multi-user.target @30.350s└─getty.target @30.350s└─getty@tty1.service @30.350s└─systemd-user-sessions.service @30.322s +2ms└─remote-fs.target @30.322s└─remote-fs-pre.target @30.322s└─nfs-client.target @30.322s└─gssproxy.service @409ms +29.912s└─basic.target @409ms└─sockets.target @409ms└─dbus.socket @409ms└─sysinit.target @408ms└─systemd-update-utmp.service @405ms +2ms└─auditd.service @381ms +24ms└─systemd-tmpfiles-setup.service @376ms +4ms└─local-fs.target @375ms└─local-fs-pre.target @299ms└─lvm2-monitor.service @140ms +158ms└─lvm2-lvmetad.service @162ms└─lvm2-lvmetad.socket @138ms└─-.slice

systemd-analyze plot Output SVG graphic showing service initialization

systemd-analyze plot > ~/plot.svg

导出的 svg 文件复制到本地电脑上就可以直接用浏览器打开看了