LPIC-2 202.1 — SysV Init and systemd

Exam weight: 3 points. You need to manage services via runlevels (SysV) and targets (systemd), understand the boot sequence, and know all the utilities listed here. The exam tests both approaches.

SysV Init

Runlevels and special levels

SysV Init divides system states into levels. Each level defines a set of running services.

Exam tip: Debian uses runlevel 2 for all multi-user modes. Red Hat uses runlevel 3 for text and 5 for graphical. This distinction is frequently tested.

Note: On-demand levels A, B, C let you run a group of scripts without changing the actual runlevel. If the current level is 2 and you run init A, the scripts for level A execute, but runlevel still shows 2.

runlevel
# Output: N 3
# N = previous level (N = initial boot), 3 = current

inittab

init is the parent of all processes. At startup it reads /etc/inittab and launches processes according to its contents, including getty for user logins.

Format of each line:

id:runlevels:action:process

• id — unique identifier (1–4 characters). For getty, id must match the tty device suffix, otherwise login accounting breaks.
• runlevels — levels without separators (e.g. 345 means levels 3, 4, and 5)
• action — what to do with the process
• process — command to run. If it starts with +, init skips utmp/wtmp accounting.

Example /etc/inittab:

# Default runlevel:
id:2:initdefault:

# System initialisation at boot (before everything else):
si::sysinit:/etc/init.d/rcS

# Single user:
~~:S:wait:/sbin/sulogin

# rc scripts for each runlevel:
l0:0:wait:/etc/init.d/rc 0
l1:1:wait:/etc/init.d/rc 1
l2:2:wait:/etc/init.d/rc 2
l3:3:wait:/etc/init.d/rc 3
l4:4:wait:/etc/init.d/rc 4
l5:5:wait:/etc/init.d/rc 5
l6:6:wait:/etc/init.d/rc 6

# Fallback on crash:
z6:6:respawn:/sbin/sulogin

# Getty on virtual terminals:
1:2345:respawn:/sbin/getty 38400 tty1
2:23:respawn:/sbin/getty 38400 tty2

action field values:

Important: Execution order: all sysinit entries run first, then boot/bootwait, then everything else.

Warning: On systemd-based systems /etc/inittab is either absent or ignored. It only applies to SysV Init.

Init scripts

Init scripts are stored in /etc/init.d/. For each runlevel there is a separate directory /etc/rcX.d/ (X = level number). When changing runlevels, init calls /etc/init.d/rc with the level number:

l2:2:wait:/etc/init.d/rc 2

The rc script then iterates over /etc/rc2.d/ and calls each symlink with the appropriate argument. Example contents of /etc/rc2.d/:

K20gpm        S11pcmcia   S20logoutd  S20ssh     S89cron
S10ipchains   S12kerneld  S20lpd      S20xfs     S91apache
S10sysklogd   S14ppp      S20makedev  S22ntpdate S99gdm

Naming convention:

• S20nginx — Start, priority 20 → calls nginx start
• K80nginx — Kill (stop), priority 80 → calls nginx stop

Scripts execute in lexicographic order. A lower number runs earlier. Start and stop are symmetric: what starts first (S20) stops last (K80).

Example init script structure:

#!/bin/sh
case "$1" in
  start)
    gpm_start
    ;;
  stop)
    gpm_stop
    ;;
  force-reload|restart)
    gpm_stop; sleep 3; gpm_start
    ;;
  *)
    echo "Usage: /etc/init.d/gpm {start|stop|restart|force-reload}"
    exit 1
esac

chkconfig

chkconfig is used on RPM-based distributions (Red Hat, CentOS, SUSE) to manage symlinks in /etc/rcX.d/.

The script header chkconfig reads:

# chkconfig: 2345 55 25
#             ^^^^ ^^ ^^
#             |    |  |--- stop priority (K-links)
#             |    |------ start priority (S-links)
#             |----------- runlevels to start on

# List all services and their runlevel status:
chkconfig --list

# List a specific service:
chkconfig --list nginx

# Enable on default runlevels (2, 3, 4, 5):
chkconfig nginx on

# Disable on all runlevels:
chkconfig nginx off

# Enable on specific levels (Red Hat syntax):
chkconfig --levels 2345 nginx on

# Add a new service (script already in /etc/init.d/):
chkconfig --add food

# Remove all runlevel links (script in /etc/init.d/ stays):
chkconfig --del food

Tip: chkconfig only manages symlinks — it does not start the service immediately. To start now, call /etc/init.d/nginx start separately.

Important: chkconfig --del food removes links from all rcX.d/ directories but leaves /etc/init.d/food on disk.

update-rc.d

Debian-based systems use update-rc.d instead of chkconfig.

# Add a service to standard runlevels:
update-rc.d foobar defaults
# Creates K-links in rc0.d, rc1.d, rc6.d and S-links in rc2.d, rc3.d, rc4.d, rc5.d

# Remove all service links (only if the script is already gone):
update-rc.d dovecot remove

# Force-remove (when the script still exists):
update-rc.d -f dovecot remove

# Create K-links (stop) on specific levels to prevent auto-start:
update-rc.d -f dovecot stop 24 2 3 4 5 .
# The trailing dot is mandatory!

Warning: If you remove links with update-rc.d -f dovecot remove, the next package update will restore them automatically. To prevent this, create K-links (stop) in the relevant directories instead.

Changing runlevel: init and telinit

# Switch runlevel immediately (no warnings):
init 3
telinit 3      # same thing via telinit

init 6         # reboot
init 0         # halt
init 1         # single-user mode

telinit is a wrapper around init that sends it a signal. On systemd systems both commands are redirected to systemctl.

Friendlier commands for multi-user systems:

shutdown -r +15 "Reboot in 15 minutes"
halt
poweroff
reboot

systemd

Units and Types

systemd manages resources through unit files. A unit defines a service or action and has the format name.type. There are eight unit types:

The exam most commonly asks about .service and .target.

Unit file structure

Example sshd.service:

[Unit]
Description=OpenSSH server daemon
After=syslog.target network.target auditd.service

[Service]
EnvironmentFile=/etc/sysconfig/sshd
ExecStartPre=/usr/sbin/sshd-keygen
ExecStart=/usr/sbin/sshd -D $OPTIONS
ExecReload=/bin/kill -HUP $MAINPID
KillMode=process
Restart=on-failure
RestartSec=42s

[Install]
WantedBy=multi-user.target

Example target file graphical.target:

[Unit]
Description=Graphical Interface
Requires=multi-user.target
After=multi-user.target
Conflicts=rescue.target
Wants=display-manager.service
AllowIsolate=yes

[Install]
Alias=default.target

Key directives:

Unit file directories

systemd reads files from three locations with different priorities:

Important: Never edit files directly in /usr/lib/systemd/system/. All customisations go in /etc/systemd/system/. Changes in /usr/lib/ will be overwritten on package updates.

Ways to override a unit:

# 1. Full replacement: create a file with the same name in /etc/systemd/system/
cp /usr/lib/systemd/system/nginx.service /etc/systemd/system/nginx.service

# 2. Partial override via a drop-in directory:
mkdir /etc/systemd/system/nginx.service.d/
cat > /etc/systemd/system/nginx.service.d/override.conf << EOF
[Service]
Restart=always
EOF

# 3. Via systemctl (creates a drop-in automatically):
systemctl edit nginx.service

# 4. Edit the full unit via systemctl:
systemctl edit --full nginx.service

# After any changes — reload configs:
systemctl daemon-reload

systemd-delta

systemd-delta shows all active unit file overrides by comparing the three directories.

systemd-delta                        # show all overrides with diffs
systemd-delta --type=overridden      # filter by type
systemd-delta --type=masked
systemd-delta --type=extended
systemd-delta --diff=false           # list only, no diff
systemd-delta --no-pager

Override types:

Targets

Targets in systemd define system states, analogous to runlevels in SysV. Key difference: multiple targets can be active simultaneously.

Dependency hierarchy: graphical.target depends on multi-user.target, which depends on basic.target.

systemctl get-default                        # current default target
systemctl set-default multi-user.target      # set default target
# Creates symlink /etc/systemd/system/default.target

systemctl list-units --type=target           # all active targets
systemctl list-unit-files --type=target      # all available targets

ls -al /etc/systemd/system/default.target
# lrwxrwxrwx ... default.target -> /usr/lib/systemd/system/graphical.target

Runlevel to target mapping

Aliases runlevel0.target through runlevel6.target exist for backwards compatibility.

systemctl

# --- Start / stop ---
systemctl start nginx.service
systemctl stop nginx.service
systemctl restart nginx.service
systemctl reload nginx.service           # reload config without restart
systemctl reload-or-restart nginx.service

# --- Status ---
systemctl status nginx.service
systemctl is-active nginx.service        # active or inactive
systemctl is-enabled nginx.service       # enabled or disabled
systemctl is-failed nginx.service

# --- Autostart ---
systemctl enable nginx.service           # add to autostart
systemctl disable nginx.service          # remove from autostart

# --- Masking ---
systemctl mask nginx.service             # symlink to /dev/null
systemctl unmask nginx.service

# --- Switching targets ---
systemctl isolate multi-user.target      # switch to multi-user (no GUI)
systemctl isolate rescue.target          # switch to rescue (no notification)
systemctl rescue                         # switch to rescue (notifies users)
systemctl rescue --no-wall               # switch without notification
systemctl emergency                      # emergency mode (broken system)
systemctl default                        # return to default target

# --- Unit information ---
systemctl list-units                     # all active units
systemctl list-units --all               # all units including inactive
systemctl list-units --type=service      # services only
systemctl list-unit-files                # all unit files with state
systemctl cat nginx.service              # print unit file contents
systemctl show nginx.service             # low-level properties
systemctl show sshd.service -p Conflicts # specific property
systemctl list-dependencies nginx.service  # dependency tree

# --- Reload systemd configuration ---
systemctl daemon-reload

Important: systemctl enable adds the service to autostart at the next boot but does not start it now. To both enable and start: run both systemctl enable and systemctl start.

Note: systemctl isolate rescue.target switches silently without notifying logged-in users. systemctl rescue first sends a wall message to all users, then switches. systemctl emergency loads a minimal environment without network services — for cases where rescue mode is already broken.

Warning: Difference between mask and disable: disable removes from autostart but the service can still be started manually. mask creates a symlink to /dev/null and completely blocks any start, including manual and via dependencies.

mkinitrd and mkinitramfs

At boot, the kernel cannot access the root filesystem if it resides on LVM, RAID, or requires a special driver. The solution is an initial ramdisk (initrd) — a temporary filesystem in memory that the kernel mounts first, containing the required modules.

mkinitrd (RPM-based distributions)

mkinitrd is used on Red Hat, CentOS, and SUSE. It automatically includes the modules needed to access the root FS, based on /etc/fstab and /etc/raidtab.

mkinitrd /boot/initrd-4.18.0-305.img 4.18.0-305.el8.x86_64

Important: Rebuild initrd every time you manually recompile the kernel or install a kernel module patch. Skipping this step means the system will fail to boot after the update.

mkinitramfs (Debian-based distributions)

mkinitramfs -o /boot/initrd.img-$(uname -r) $(uname -r)

# Options:
mkinitramfs -d /etc/initramfs-tools -o outfile   # alternative confdir
mkinitramfs -k -o outfile                         # keep temp directory
mkinitramfs -r /dev/sda1 -o outfile               # override ROOT from config

Config file: /etc/initramfs-tools/initramfs.conf

Warning: On Debian-based distributions mkinitrd is broken for modern kernels. Use mkinitramfs only.

Root Device Configuration

The root device can be set in several ways, each overriding the previous:

1. Defaults from kernel source code
2. Values set by the rdev command
3. Parameters passed to the kernel at boot (root=/dev/xyz)
4. Parameters in the GRUB configuration file

From within the initrd environment, the root device can be changed via /proc:

# Change root device to /dev/hda1 (0x301):
echo 0x301 > /proc/sys/kernel/real-root-dev

# Configure NFS root:
echo /var/nfsroot > /proc/sys/kernel/nfs-root-name
echo 193.8.232.2:193.8.232.7::255.255.255.0:myhost > /proc/sys/kernel/nfs-root-addrs
echo 255 > /proc/sys/kernel/real-root-dev   # 0xff = NFS

Linux Boot Process

The kernel tries to launch init from the following paths in order until one succeeds:

/sbin/init
/etc/init
/bin/init
/bin/sh

Important: If none of these paths work, the kernel panics with Kernel panic - not syncing. The exam may ask which path is tried first and what happens on failure.

Linux Standard Base (LSB)

LSB defines a compatibility standard between Linux distributions. A program built in an LSB-compliant environment will run on any LSB-supporting distribution.

What LSB defines:

• Standard libraries: libdl, libcrypt, libpthread, libc, libm, including search paths, format (ELF), and dynamic linking
• 130+ commands with calling conventions: cp, tar, kill, gzip, perl, python
• System init behaviour, init script functions, and locations
• Runlevel definitions, username/group conventions, and UID/GID ranges
• Package standard: LSB uses RPM format. On Debian, alien reads RPM packages

Note: LSB is a family of specifications, not a single one. There is a generic part (LSB-generic) and an architecture-specific part (LSB-arch). A binary for Intel will not run on Alpha even if both distributions support LSB.

LSB init scripts

An LSB-compliant init script must support these arguments:

Functions from /lib/lsb/init-functions:

# Start a program as a daemon (checks if already running):
start_daemon [-f] [-n nicelevel] [-p pidfile] pathname [args...]
# -f: start even if already running
# -n: set nice level

# Stop a process (tries specified signal, falls back to SIGTERM):
killproc [-p pidfile] pathname [signal]

# Get the PID(s) of a daemon:
pidofproc [-p pidfile] pathname

Example LSB header in a script:

### BEGIN INIT INFO
# Provides:          myservice
# Required-Start:    $network $remote_fs $syslog
# Required-Stop:     $network $remote_fs $syslog
# Default-Start:     2 3 4 5
# Default-Stop:      0 1 6
# Short-Description: My custom service
# Description:       Long description of the service
### END INIT INFO

Tip: Required-Start defines dependencies. $network, $remote_fs, $syslog are LSB virtual facilities. Both chkconfig and update-rc.d read this header to create symlinks with the correct priority numbers.

Exam Cheat Sheet

Paths and Files

Key Commands

Common Mistakes

• Forgetting systemctl daemon-reload after editing a unit file
• Confusing enable (autostart) with start (start now)
• Confusing disable (removes autostart, manual start still works) with mask (blocks everything)
• Editing files directly in /usr/lib/systemd/system/
• Forgetting the trailing dot in update-rc.d -f dovecot stop 24 2 3 4 5 .
• Confusing Red Hat and Debian runlevel behaviour (levels 2–5)

Practice Questions

Q1. On a Debian server with SysV Init, you need to disable autostart of apache2. Which command does this?

A. chkconfig apache2 off B. update-rc.d -f apache2 remove C. systemctl disable apache2 D. service apache2 stop

Answer: B. chkconfig is for RPM-based systems. update-rc.d -f apache2 remove removes symlinks on Debian. systemctl disable only works with systemd.

Q2. Which file stores the default runlevel in SysV Init?

A. /etc/rc.local B. /etc/default/init C. /etc/inittab D. /etc/init.d/defaults

Answer: C. The line id:3:initdefault: in /etc/inittab sets the boot runlevel.

Q3. An admin ran systemctl edit nginx.service, saved changes, but the service behaviour is unchanged. What needs to be done?

A. systemctl restart nginx.service B. systemctl reload nginx.service C. systemctl daemon-reload D. systemctl reset nginx.service

Answer: C. After changing unit files, systemd must reload its configuration with systemctl daemon-reload. Only then do the changes take effect.

Q4. What happens to files in /run/systemd/system/ on reboot?

A. They are copied to /etc/systemd/system/ B. They are preserved unchanged C. They are deleted D. They are moved to /usr/lib/systemd/system/

Answer: C. The /run/ directory holds temporary data that does not survive a reboot. Override files there are discarded.

Q5. Which command shows which unit files are overridden on the system?

A. systemctl list-overrides B. systemd-delta C. systemctl diff D. systemctl show --changed

Answer: B. systemd-delta compares all three unit file directories and shows discrepancies with type labels: masked, overridden, extended, etc.

Q6. An admin wants myapp.service to never start, even if triggered manually or via a dependency. What should they do?

A. systemctl disable myapp.service B. systemctl stop myapp.service C. systemctl mask myapp.service D. systemctl remove myapp.service

Answer: C. mask creates a symlink to /dev/null, completely blocking any start. disable only removes from autostart.

Q7. Which systemd target corresponds to SysV runlevel 5 on Red Hat systems?

A. multi-user.target B. rescue.target C. graphical.target D. runlevel5.target

Answer: C. Runlevel 5 on Red Hat is multi-user graphical mode, which maps to graphical.target. runlevel5.target is an alias pointing to graphical.target.

Q8. What does systemctl isolate multi-user.target do?

A. Starts all units from multi-user.target without touching others B. Starts units from multi-user.target and stops all units outside its dependency tree C. Sets multi-user.target as the default target D. Lists units in multi-user.target

Answer: B. isolate switches the system to the specified target: starts everything needed and stops everything else. It is the equivalent of changing runlevel in SysV.