SystemV vs systemd

SystemV vs systemd Slide 1 of 60
SystemV vs systemd
All Things Open
October, 2015
This presentation is taken from my class
“Theory and Practice of Linux System Administration”

Legal Stuff
● Linux is a registered trademark of Linus Torvalds
● Red Hat is a registered trademark of Red Hat, Inc.
● Unix is a registered trademark of AT&T
● Other marks are the trademarks of their respective
owners
Copyright©
2015 David P. Both
Content licensed under the terms of the Creative Commons Attribution-ShareAlike 4.0 International
license ("CC-BY-SA")

David Both
dboth@millennium-technology.com
David.Both@OpenSource.com
RHCE, SCSA

Introducing systemd
● Developed by
● Lennart Poettering
● Kay Sievers
● Objectives
● Fix outdated SystemV startup
● Start less
● Only daemons that are actually needed
● Start more in parallel
● Improve support for hotplug hardware
● Maintain compatibility with SystemV start scripts

systemd Functions
● A system and service manager (manages
both the system and its services)
● A software platform (serves as a basis for
developing other software)
● The glue between applications and the
kernel (provides various interfaces that
expose functionalities provided by the
kernel)

Strategy
● At least 2 ways to ensure that a service is
available when it is needed
● Keep track of all other services which may
need it and be sure to start things in the right
order
● Wait until some task tries to connect to the
service and start it on demand
● systemd takes the second approach
● Similar to MacOS launchd

Implementation
● Create sockets for services
● Serialization during startup was due to
waiting for dependent services sockets to be
created
● When a connection request arrives on a
specific socket, the associated daemon is
started
● Detailed systemd information
● http://0pointer.de/blog/projects/systemd.html
● http://www.freedesktop.org/wiki/Software/systemd

systemd by Release
● Started showing up in about Fedora 14
● Available as alternative
● Fedora 15
● First release with systemd as default
for many services
● Conversion essentially complete by
Fedora 17
● CentOS 7

Benefits
● Speed startup
● Start as much in parallel as possible
● Only start what needs to be started
● Manage dynamic hardware hotplug
● Static hardware managed like dynamic

systemd Pros and Cons
● Not portable
● Only works in Linux
● Opportunity to do things better for Linux
● More binary executables
● Faster
● Fewer shell scripts
● Less discoverable
● Open Source code
● ASCII Text configuration files
● Still provides openness

Linux Boot Sequence

BIOS POST
• Checks basic operability of hardware
• INT 13H Locates Boot Sector
• Boot Loader Stage 1
– GRUB
– LILO

GRUB
• Stage 1 Loads GRUB Stage 1.5
– File system specific
• Loads Stage 2
• /boot/grub/grub.conf

GRUB
● GRUB locates the kernel image and loads it
● Located in /boot
● Standard EXT3/4 partition
● Cannot be LVM
● Located at beginning of disk
● Usually loads an initrd image
● Select from multiple kernels

grub.conf
●
Defines grub menu options
– Allows selection of boot kernel
– Edit kernel boot parameters
●
Specifies kernel boot parameters
●
Boot hard drive

GRUB2
• Introduced in Fedora 16
• Command based pre-OS environment
• GRUB2 Locates the Kernel image and loads it
– Located in /boot/grub2
• Standard ext3/4 partition
• Cannot be LVM
• Located at beginning of disk
– May also load initrd image
– Multiple kernel

Kernel
• Uncompresses itself into RAM
• Loads device drivers
• Mounts /
• Launches init or systemd

Kernel Boot Messages
● Ring buffer
● Console
● dmesg command
● /var/log/messages
● grep kernel:
● /var/log/dmesg
● Discontinued with systemd
● Never used by many distros
● Timestamps in square braces
● Recent releases of Fedora

Sample DMESG Log
[ 0.000000] NR_IRQS:2304
[ 0.000000] CPU 0 irqstacks, hard=f3c18000 soft=f3c1a000
[ 0.000000] Console: colour VGA+ 80x25
[ 0.000000] console [tty0] enabled
[ 0.000000] Fast TSC calibration failed
[ 0.000000] TSC: Unable to calibrate against PIT
[ 0.000000] TSC: HPET/PMTIMER calibration failed.
[ 0.000000] Marking TSC unstable due to could not calculate TSC khz
[ 0.016998] Calibrating delay loop... 2027.52 BogoMIPS (lpj=1013760)
[ 0.038997] pid_max: default: 32768 minimum: 301
[ 0.043996] Security Framework initialized
[ 0.043996] SELinux: Initializing.
[ 0.044996] SELinux: Starting in permissive mode
[ 0.049996] Mount-cache hash table entries: 512
[ 0.076993] Initializing cgroup subsys ns
[ 0.077993] ns_cgroup deprecated: consider using the 'clone_children'
flag without the ns_cgroup.
[ 0.077993] Initializing cgroup subsys cpuacct
[ 0.079993] Initializing cgroup subsys memory
[ 0.081992] Initializing cgroup subsys devices
[ 0.081992] Initializing cgroup subsys freezer
[ 0.082992] Initializing cgroup subsys net_cls
[ 0.082992] Initializing cgroup subsys blkio
[ 0.086992] mce: CPU supports 0 MCE banks

Devices
● Device files located in /dev
● Previously all possible device files created at
installation
● Tens of thousands
● Most unused
● NOT device drivers
● Device file major/minor codes
● Assigned by Linux assigned names and numbers authority
(LANANA)
● Current device list can be found on web site

Devices
● Enter udev
● Creates only as required – mostly
● Treats all devices as plug'n'pray
● Reduces /dev/chaos
● udev rules

DBUS
● Inter-Process Communications (IPC)
● Daemon for kernel
● User space for applications
● New devices plugged in
● Triggers udev to create a new device file
● Application coordination
● Manage communications
● Multiple instances of one application
● Multiple different applications
● DBUS rules

Linux Startup

Boot vs Startup
●
Boot ends with loading the kernel and init or
systemd
●
Startup begins when the init or systemd process
takes control
●
init
●
The entire startup process is handled by scripts
●
systemd
●
Binary executables
●
Configuration files
●
Compatible with SystemV init scripts

Upstart
●
Was to have replaced traditional start scripts
and /etc/inittab
●
/etc/init
●
Changeover started in some distros
– Fedora
– CentOS
– Ubuntu
●
Quickly superseded by systemd

Traditional Startup
Using init and SystemV Start Scripts

/sbin/init
● ELF Binary
● Controls Linux initialization and runlevel startup
● Mother of all processes
● PID #1

Startup Files
• Configuration files
– /etc/inittab
– /etc/fstab
• Log Files
– /var/log/dmesg
– /var/log/boot.log

/etc/inittab
• Defines runlevels 0 through 6
• Specifies default runlevel
• gettys for defined runlevels
• Ctrl-Alt-Del action
• UPS power failure actions
– Rarely used

inittab 1
#
# inittab This file describes how the INIT process should set up
# the system in a certain run-level.
#
# Author: Miquel van Smoorenburg, <miquels@drinkel.nl.mugnet.org>
# Modified for RHS Linux by Marc Ewing and Donnie Barnes
#
# Default runlevel. The runlevels used by RHS are:
# 0 - halt (Do NOT set initdefault to this)
# 1 - Single user mode
# 2 - Multiuser, without NFS (The same as 3, if you do not have
networking)
# 3 - Full multiuser mode
# 4 - unused
# 5 - X11
# 6 - reboot (Do NOT set initdefault to this)

inittab 2
#
id:5:initdefault:
# System initialization.
si::sysinit:/etc/rc.d/rc.sysinit
l0:0:wait:/etc/rc.d/rc 0
# Trap CTRL-ALT-DELETE
ca::ctrlaltdel:/sbin/shutdown -t3 -r now
●

inittab 3
# When our UPS tells us power has failed, assume we have a few minutes
# of power left. Schedule a shutdown for 2 minutes from now.
# This does, of course, assume you have powerd installed and your
# UPS connected and working correctly.
pf::powerfail:/sbin/shutdown -f -h +2 "Power Failure; System Shutting
Down"
# If power was restored before the shutdown kicked in, cancel it.
pr:12345:powerokwait:/sbin/shutdown -c "Power Restored; Shutdown
Cancelled"
# Run gettys in standard runlevels
1:2345:respawn:/sbin/mingetty tty1
# Run xdm in runlevel 5
x:5:respawn:/etc/X11/prefdm -nodaemon

/etc/fstab
●
Common to all startup systems
●
Specifies disk partitions to mount
●
Some directories must be part of / (root)
– /etc
– /dev
– /bin
– /sbin
– /lib
– /usr
• For only a couple releases of Fedora

Typical fstab
#
# /etc/fstab
# Created by anaconda on Thu Jun 9 15:15:28 2011
#
# Accessible filesystems, by reference, are maintained under '/dev/disk'
# See man pages fstab(5), findfs(8), mount(8) and/or blkid(8) for more info
#
/dev/mapper/vg_instructor-root / ext4 defaults 1 1
UUID=d1a58f39-5002-4150-933b-131fecb106c5 /boot ext4 defaults 1 2
/dev/mapper/vg_instructor-home /home ext4 defaults 1 2
/dev/mapper/vg_instructor-tmp /tmp btrfs defaults 1 2
/dev/mapper/vg_instructor-usr /usr ext4 defaults 1 2
/dev/mapper/vg_instructor-var /var ext4 defaults 1 2
/dev/mapper/vg_instructor-swap swap swap defaults 0 0
tmpfs /dev/shm tmpfs defaults 0 0
devpts /dev/pts devpts gid=5,mode=620 0 0
sysfs /sys sysfs defaults 0 0
proc /proc proc defaults 0 0

Another fstab
/dev/mapper/vg_david1-root       /                   ext4    defaults        1 1
/dev/mapper/vg_david2-Virtual    /Virtual            ext4    defaults        1 2
#UUID=d352f5a0-f399-4d9c-b46f-02dd81dca3f8 /boot     ext4    defaults        1 2
#/dev/sda1                       /boot     ext4    defaults        1 2
LABEL=/boot                      /boot               ext4    defaults        1 2
/dev/mapper/vg_david2-home       /home               ext4    defaults        1 2
/dev/mapper/vg_david2-stuff      /stuff              ext4    defaults        1 2
/dev/mapper/vg_david1-tmp        /tmp                ext4    defaults        1 2
/dev/mapper/vg_david1-usr        /usr                ext4    defaults        1 2
/dev/mapper/vg_david1-usrlocal   /usr/local          ext4    defaults        1 2
/dev/mapper/vg_david2-var        /var                ext4    defaults        1 2
/dev/mapper/vg_david1-swap       swap                swap    defaults        0 0
# Various USB HDDs
# Using labels makes it possible to mount at the same location consistently
LABEL=WD-500GB-USB      /media/WD-500GB-USB     ext4   auto,owner,user,defaults 0 0
LABEL=Iomega-1TB-USB    /media/Iomega-1TB-USB   ext4   auto,owner,user,defaults 0 0
LABEL=Iomega-160GB-USB  /media/Iomega-160GB-USB ext4   auto,owner,user,defaults 0 0
tmpfs                   /dev/shm                tmpfs defaults        0 0
devpts                  /dev/pts                devpts gid=5,mode=620 0 0
sysfs                   /sys                    sysfs defaults        0 0
proc                    /proc                   proc   defaults        0 0

/etc/rc.d/rc.sysinit
• Bash script started by init
• Manages other startup processes
– Set hostname
– Start SELINUX
– Set the clock
– Load modular kernel device driver modules
• /etc/modprobe.conf
– Remount / as rw
– Mount other file systems
– Sets quotas
– etc...

/etc/rc.d/rc
• Bash script
• Started by init after rc.sysinit ends
• Changes runlevels
• rc scripts

Start Scripts
• /etc/rc.d/init.d
• /etc/rcX.d
– Links
– K = Kill
– S = Start
• service command
– service <service name> start|stop|
restart|status|reload ...

SystemV Runlevels
0 Poweroff
1,S,s Single User
2 Multi-user, no NFS
3 Normal, multi-user with NFS
4 Unused but basically same as 3
5 GUI
6 Reboot

Runlevels
• /etc/rc.d/rc
– Script to change runlevels
– Kill scripts first
– Then Start scripts
• rc scripts in /etc/init.d
• K or S links in /etc/rc.d/rcX.d
– Links to directories in /etc/init.d

Runlevel Management
SystemV
• init or telinit
● Change runlevel now
• chkconfig
● For persistent configuration
• The service command
● Turn on/off now
● Default runlevel
● /etc/inittab

Startup
Using systemd

systemd Units
● Service
● Most obvious kind of unit
● Daemons that can be started, stopped, restarted, reloaded
● For compatibility can also read classic SysV init
scripts
● In particular parse the LSB header, if it exists
● /etc/init.d is just another source of configuration

systemd Units
● Socket
● Encapsulates a socket in the file-system or on
the Internet
● Currently support AF_INET, AF_INET6, AF_UNIX
sockets of the types stream, datagram, and sequential
packet
● Classic FIFOs as transport

systemd Units
● Each socket unit has a matching service unit
● Service started when the first connection comes in on the
socket or FIFO
● Example: nscd.socket starts nscd.service on an incoming
connection
● Sockets can be viewed with netstat command

systemd Units
● Device
● Encapsulates a device in the Linux device tree
● If a device is marked for this via udev rules, it
will be exposed as a device unit in systemd
● Properties set with udev can be used as
configuration source to set dependencies for
device units

systemd Units
● Mount
● Encapsulates a mount point in the file system
hierarchy
● systemd monitors all mount points, how they come and
go, and can also be used to mount or unmount mount-
points
● /etc/fstab used as an additional configuration
source for mount points
● similar to how SysV init scripts can be used as additional
configuration source for service units.

systemd Units
● Automount
● Encapsulates an automount point in the file
system hierarchy
● Each automount unit has a matching mount unit,
which is started (i.e. mounted) as soon as the
automount directory is accessed

systemd Units
● Target
● Logical grouping of units
● Instead of actually doing anything by itself it simply
references other units, which are controlled together
● Examples for this are:
● multi-user.target equivalent to run-level 3 on classic SysV system
● bluetooth.target pulls in bluetooth related services that otherwise would not
need to be started: bluetoothd and obexd, etc.
● graphical.target depends upon multi-user.target

systemd Units
● Snapshot
● Similar to target units
● Snapshots do not actually do anything themselves
● Purpose is to reference other units
● Save/rollback the state of services and units of the init
system
● Two intended use cases
● Allow the user to temporarily enter a specific state such as
"Emergency Shell"
● Ease support for system suspending

Control Groups
● cgroups define groups of processes
● Process aggregation
● Control of cgroups
● Resource allocation
● Runaway process spawn
● /cgroup directory
● Let me know if you find anything here
● /sys/fs/cgroup/systemd/
● /proc/<PID>/cgroup

systemd
● Configuration
● /etc/systemd
● Modify here
● /lib/systemd
● Don't change this
● Management
● systemctl
● init command is softlink to systemd
● The service command still works for many
services
● Redirected through systemd

systemctl Command
● List running units
● systemctl
● List all units
● systemctl -a list-units
● Start and stop
● systemctl start|stop <target>
● Display device status
● systemctl status udisks2.service
● Disable startup of a service
● systemctl disable telnet.service

systemctl
systemctl -a
systemctl stop|start cups.service

cgroups
● Cgroup tree
● systemd-cgls
● System slice
● User slice
● PS with Cgroups
● ps xawf -eo pid,user,cgroup,args

systemd Targets
● Default
● Symlink to the desired run target
● /etc/systemd/system/default.target
● multiuser
● multi-user.target
● graphical
● graphical.target
● Still has target runlevels 0-6
● /lib/systemd/system
● Runlevels 2,3,4 all point to multi-user.target

systemd Targets
● Single User
● Recovery mode
● Somewhat more functionality than SystemV runlevel 1 or
Single User mode

Target related commands
● List all targets
● systemctl listunits –type=target
● Set default target
● systemctl setdefault <name of
target>.target
● Display current default target
● systemctl getdefault

Changing the Default Target
● systemctl setdefault <name of
target>.target
● ln sf /lib/systemd/system/runlevel5.target
/etc/systemd/system/default.target
● ln sf /lib/systemd/system/multi
user.target
● ln sf /lib/systemd/system/graphical.target

Thank you

SystemV vs systemd

Recommandé

Recommandé

Contenu connexe

Tendances

Tendances (20)

En vedette

En vedette (20)

Similaire à SystemV vs systemd

Similaire à SystemV vs systemd (20)

Plus de All Things Open

Plus de All Things Open (20)

Dernier

Dernier (20)

SystemV vs systemd