• https://freedesktop.org/wiki/Software/systemd/Debugging/

• journalctl -u <SERVICE>

  • -b<INTEGER> log from last boot. That is, 0 "current", 1 "previous", ... .
  • -e show logs from the last entry.
  • -f show the most recent entries and monitor output for new ones.

• systemd-analyze verify <FILE> ensure that the service file is valid.

• systemctl list-unit-files --type=service --state=enabled

• systemctl edit <SERVICE>
• Add new section [SERVICE] and define thereunder Environment=SYSTEMD_LOG_LEVEL=debug
• Restart service systemctl restart <SERVICE>
• Revert changes systemctl revert <SERVICE>
• systemctl restart <SERVICE>

Control the system locale and keyboard layout settings.

• localectl set-x11-keymap <layout>

• Part of SystemD
• Only EFI boot

1. Verify if system is in UEFI mode by confirming presence of
   • efivar --list or ls /sys/firmware/efi/efivars.
2. Mount the boot ESP drive
3. bootctl install
   • Creates the EFI and loader directories.

Example of boot as the install directory:

/boot/
+-- EFI/
+-- loader/
|   +-- entries/
|   |   +-- archlinux.conf
|   |   +-- archlinux-fallback.conf
|   +-- keys/
|   +-- entries.srel
|   +-- loader.conf
|   +-- random-seed
+-- initramfs-linux.img
+-- initramfs-linux.img
+-- initramfs-linux-fallback.img
+-- vmlinuz-linux

• As of systemd v257 bootctl does not create the UEFI boot entry in NVRAM when running in pid namespace, for example as is the case for arch-chroo.

A work around is to install the boot loader from the non-chroot environment:

• AdelKS on [bootctl] Regression: Silent failure to create EFI boot entry from within chroot #36174

bootctl --esp-path=/mnt/boot install
arch-chroot /mnt
bootctl install

Note # If ESP path is mounted to a sub-directory then entry conf had problem finding the initramfs and vmlinuz.

• /boot/esp as mount point
• Next time, try ../vmlinuz-linux

• systemd-networkd enables systemd-networkd-wait-online.service
• Create configuration files in /etc/systemd/network/
• Overview networkctl list
• udevadm test-builtin net_setup_link /sys/class/net/<NAME OF INTERFACE TO TEST>
• Create a WPA Supplicant for the wireless device.

Delete a started device (link): networkctl delete <LINK>

Example configuration:

[Match]
Name=device_ap
WLANInterfaceType=ap

[Link]
RequiredForOnline=routable

[Network]
ConfigureWithoutCarrier=false

DHCPServer=true

IPv4Forwarding=true

Address=192.168.2.2/24

[DHCPServer]
ServerAddress=192.168.2.2/24

PoolOffset=2
PoolSize=20
EmitDNS=yes
DNS=1.1.1.1 4.4.4.4 8.8.8.8 ...

The device is connected to the AP and responds to ping. From the device:

• ping an IP address if it fails then the problem is routing, NAT or firewall
• ping a domain name if it fails then the problem is DNS

SystemD NetworkD can handle IP forwarding instead of DHCPD

• Start the DHCP Server DHCPServer=yes under the section [Network]
• [DHCPServer] to configure, for example:
  • Poolsize=20

Clean DHCP server leases

• /var/lib/systemd/network/dhcp-server-lease/<device name>

• Debug tcpdump -i enp1s0 -nn -s0 -v -A udp port 67
• Can systemd networkd be configured for netboot, PXE boot, if yes, how?

[Match]
Name=<NETWORK INTERFACE>

[Link]
RequiredForOnline=False

[Network]
DHCP=yes
DHCPServer=true

IPForward=yes
IPMasquerade=both

Address=10.0.0.2/24

ConfigureWithoutCarrier=true

[DHCPServer]
PoolOffset=100
PoolSize=1

BootServerAddress=10.0.0.2

# 12 "Hostname"
# Hostname=
SendOption=12:string:<CLIENT HOSTNAME>

# 17 "Root Path"
SendOption=17:string:/srv/netboot

# EFI
#SendOption=67:string:netboot/x86_64-efi/core.efi
#BootFilename=/srv/netboot/grub/x86_64-efi/core.efi
BootFilename=grub/x86_64-efi/core.efi

# i386
#SendOption=67:string:netboot/grub/i396-pc/core.0
#BootFilename=netboot/grub/i396-pc/core.0
#BootFilename=/srv/tftp/netboot/grub/i396-pc/core.0

[DHCPServerStaticLease]
# Device #1
MACAddress=1a:2b:3c:4d:5e:6f
Address=10.0.0.101

#[DHCPServerStaticLease]
# Device #2
MACAddress=a1:b2:c3:d4:e5:f6
#Address=10.0.0.102

SystemD NetworkD can handle IP forwarding instead of /etc/sysctl.d/30-ipforward.conf

• IPForward=yes under the section [Network]

Create a link configuration that lexically comes before "99". For example, 00-mac-wlv.link.

Example, match virtual devices.

• Could not use MACAddress because "MAC address on the device already set by userspace.".
• The #PermanentMACAddress matches the physical device, and from it all derived virtual interfaces.

• Wireless example

The order of the files for devices in /etc/systemd/network/ has not been tested. The following worked:

• 00-mac-ether.link
• 01-mac-wireless.link
• 10-wan-ether.network
• 11-lan-ether.network
• 20-wan-wireless.network
• 30-lan-wireless-ap.netdev
• 31-lan-wireless-ap.network

Address RouteMetric applies to connected route
Route   Metric deals with static route
DHCPxx  RouteMetric applies to routes learned from DHCP

• Arch Linux Wiki WireGuard
• 40-wan-vpn-wg.netdev
  • If the private key is provided in the configuration file or another file the permission of that file must be
    • chown root:systemd-network
    • chmod 0640
  • Note: changes to the content of the configuration files may require a restart of systemd-networkd instead of networkctl reload.
• 41-wan-vpn-wg.network

Create a configuration

• /etc/wpa_supplicant/wpa_supplicant-wlp0s0.conf

• /etc/wpa_supplicant/wpa_supplicant-nl80211-wlp0s0.conf

• wpa_supplicant-<interface>.conf

• /etc/systemd/network/10-wlp0s0.network

• systemctl enable --now [email protected]

• systemctl enable --now [email protected]

• https://www.ntppool.org/en/

/etc/systemd/timesyncd.conf

systemctl status systemd-timesyncd systemctl start/restart systemd-timesyncd

timedatectl show-timesync --all

Display current system time.

• # timedatectl status

Enable automatic NTP synchronization.

• # timedatectl set-ntp true

Force NTP synchronization.

• # systemctl restart systemd-timesyncd

• /etc/system/system/<UNIT>
• systemd-analyze plot > overview.svg
• systemctl reset-failed fix references to missing unit files

• When the storage device in inserted then auto-mount it
• When removing the device then auto-unmount it

Two methods:

• x-systemd.automount - on failure, do nothing
• unit.mount, unit.automount - on failure, raise error

When accessing the mount point and the device is not available then content of the mount point is accessed, usually an empty directory, i.e. no error is raised.

For example, in fstab:

UUID=<UUID>  /<MOUNT-POINT> ext4  rw,noatime,discard,noauto,x-systemd.automount  0 1

In order for the change to take effect:

• sudo systemctl daemon-reload
• sudo systemctl restart remote-fs.target
• sudo systemctl restart local-fs.target

When accessing the mount point and the device is not available then an error is raised.

1. Create a unit file in /etc/systemd/user/<UNIT>
   • with ExecStart and ExecStop
   • If the execution requires root privileges then execute with =+ "full privileges". For example:
     • ExecStart=+/usr/bin/bash -c "echo 'heartbeat' > /sys/class/leds/red:power/trigger
2. Enable the file globally systemctl --global enable <unit>

Note that the global unit file cannot have the same name as a system unit file.

List global unit files.

• systemctl --global list-unit-files