systemd systemd.link 5 systemd.link Network device configuration link.link A plain ini-style text file that encodes configuration for matching network devices, used by systemd-udevd8 and in particular its net_setup_link builtin. See systemd.syntax7 for a general description of the syntax. The link files are read from the files located in the system network directory /usr/lib/systemd/network, the volatile runtime network directory /run/systemd/network, and the local administration network directory /etc/systemd/network. Link files must have the extension .link; other extensions are ignored. All link files are collectively sorted and processed in lexical order, regardless of the directories in which they live. However, files with identical filenames replace each other. Files in /etc/ have the highest priority, files in /run/ take precedence over files with the same name in /usr/lib/. This can be used to override a system-supplied link file with a local file if needed. As a special case, an empty file (file size 0) or symlink with the same name pointing to /dev/null disables the configuration file entirely (it is "masked"). Along with the link file foo.link, a "drop-in" directory foo.link.d/ may exist. All files with the suffix .conf from this directory will be parsed after the file itself is parsed. This is useful to alter or add configuration settings, without having to modify the main configuration file. Each drop-in file must have appropriate section headers. In addition to /etc/systemd/network, drop-in .d directories can be placed in /usr/lib/systemd/network or /run/systemd/network directories. Drop-in files in /etc/ take precedence over those in /run/ which in turn take precedence over those in /usr/lib/. Drop-in files under any of these directories take precedence over the main link file wherever located. The link file contains a [Match] section, which determines if a given link file may be applied to a given device, as well as a [Link] section specifying how the device should be configured. The first (in lexical order) of the link files that matches a given device is applied. Note that a default file 99-default.link is shipped by the system. Any user-supplied .link should hence have a lexically earlier name to be considered at all. See udevadm8 for diagnosing problems with .link files. A link file is said to match a device if all matches specified by the [Match] section are satisfied. When a link file does not contain valid settings in [Match] section, then the file will match all devices and systemd-udevd warns about that. Hint: to avoid the warning and to make it clear that all interfaces shall be matched, add the following: OriginalName=* The following keys are accepted: MACAddress= A whitespace-separated list of hardware addresses. Use full colon-, hyphen- or dot-delimited hexadecimal. See the example below. This option may appear more than once, in which case the lists are merged. If the empty string is assigned to this option, the list of hardware addresses defined prior to this is reset. Example: MACAddress=01:23:45:67:89:ab 00-11-22-33-44-55 AABB.CCDD.EEFF PermanentMACAddress= A whitespace-separated list of hardware's permanent addresses. While MACAddress= matches the device's current MAC address, this matches the device's permanent MAC address, which may be different from the current one. Use full colon-, hyphen- or dot-delimited hexadecimal. This option may appear more than once, in which case the lists are merged. If the empty string is assigned to this option, the list of hardware addresses defined prior to this is reset. Path= A whitespace-separated list of shell-style globs matching the persistent path, as exposed by the udev property ID_PATH. Driver= A whitespace-separated list of shell-style globs matching the driver currently bound to the device, as exposed by the udev property ID_NET_DRIVER of its parent device, or if that is not set, the driver as exposed by ethtool -i of the device itself. If the list is prefixed with a "!", the test is inverted. Type= A whitespace-separated list of shell-style globs matching the device type, as exposed by networkctl status. If the list is prefixed with a "!", the test is inverted. Property= A whitespace-separated list of udev property name with its value after a equal (=). If multiple properties are specified, the test results are ANDed. If the list is prefixed with a "!", the test is inverted. If a value contains white spaces, then please quote whole key and value pair. If a value contains quotation, then please escape the quotation with \. Example: if a .link file has the following: Property=ID_MODEL_ID=9999 "ID_VENDOR_FROM_DATABASE=vendor name" "KEY=with \"quotation\"" then, the .link file matches only when an interface has all the above three properties. OriginalName= A whitespace-separated list of shell-style globs matching the device name, as exposed by the udev property "INTERFACE". This cannot be used to match on names that have already been changed from userspace. Caution is advised when matching on kernel-assigned names, as they are known to be unstable between reboots. Host= Matches against the hostname or machine ID of the host. See ConditionHost= in systemd.unit5 for details. When prefixed with an exclamation mark (!), the result is negated. If an empty string is assigned, then previously assigned value is cleared. Virtualization= Checks whether the system is executed in a virtualized environment and optionally test whether it is a specific implementation. See ConditionVirtualization= in systemd.unit5 for details. When prefixed with an exclamation mark (!), the result is negated. If an empty string is assigned, then previously assigned value is cleared. KernelCommandLine= Checks whether a specific kernel command line option is set. See ConditionKernelCommandLine= in systemd.unit5 for details. When prefixed with an exclamation mark (!), the result is negated. If an empty string is assigned, then previously assigned value is cleared. KernelVersion= Checks whether the kernel version (as reported by uname -r) matches a certain expression. See ConditionKernelVersion= in systemd.unit5 for details. When prefixed with an exclamation mark (!), the result is negated. If an empty string is assigned, then previously assigned value is cleared. Architecture= Checks whether the system is running on a specific architecture. See ConditionArchitecture= in systemd.unit5 for details. When prefixed with an exclamation mark (!), the result is negated. If an empty string is assigned, then previously assigned value is cleared. The [Link] section accepts the following keys: Description= A description of the device. Alias= The ifalias interface property is set to this value. MACAddressPolicy= The policy by which the MAC address should be set. The available policies are: persistent If the hardware has a persistent MAC address, as most hardware should, and if it is used by the kernel, nothing is done. Otherwise, a new MAC address is generated which is guaranteed to be the same on every boot for the given machine and the given device, but which is otherwise random. This feature depends on ID_NET_NAME_* properties to exist for the link. On hardware where these properties are not set, the generation of a persistent MAC address will fail. random If the kernel is using a random MAC address, nothing is done. Otherwise, a new address is randomly generated each time the device appears, typically at boot. Either way, the random address will have the unicast and locally administered bits set. none Keeps the MAC address assigned by the kernel. Or use the MAC address specified in MACAddress=. An empty string assignment is equivalent to setting none. MACAddress= The interface MAC address to use. For this setting to take effect, MACAddressPolicy= must either be unset, empty, or none. NamePolicy= An ordered, space-separated list of policies by which the interface name should be set. NamePolicy= may be disabled by specifying net.ifnames=0 on the kernel command line. Each of the policies may fail, and the first successful one is used. The name is not set directly, but is exported to udev as the property ID_NET_NAME, which is, by default, used by a udev7, rule to set NAME. The available policies are: kernel If the kernel claims that the name it has set for a device is predictable, then no renaming is performed. database The name is set based on entries in the udev's Hardware Database with the key ID_NET_NAME_FROM_DATABASE. onboard The name is set based on information given by the firmware for on-board devices, as exported by the udev property ID_NET_NAME_ONBOARD. See systemd.net-naming-scheme7. slot The name is set based on information given by the firmware for hot-plug devices, as exported by the udev property ID_NET_NAME_SLOT. See systemd.net-naming-scheme7. path The name is set based on the device's physical location, as exported by the udev property ID_NET_NAME_PATH. See systemd.net-naming-scheme7. mac The name is set based on the device's persistent MAC address, as exported by the udev property ID_NET_NAME_MAC. See systemd.net-naming-scheme7. keep If the device already had a name given by userspace (as part of creation of the device or a rename), keep it. Name= The interface name to use. This option has lower precedence than NamePolicy=, so for this setting to take effect, NamePolicy= must either be unset, empty, disabled, or all policies configured there must fail. Also see the example below with Name=dmz0. Note that specifying a name that the kernel might use for another interface (for example eth0) is dangerous because the name assignment done by udev will race with the assignment done by the kernel, and only one interface may use the name. Depending on the order of operations, either udev or the kernel will win, making the naming unpredictable. It is best to use some different prefix, for example internal0/external0 or lan0/lan1/lan3. AlternativeNamesPolicy= A space-separated list of policies by which the interface's alternative names should be set. Each of the policies may fail, and all successful policies are used. The available policies are database, onboard, slot, path, and mac. If the kernel does not support the alternative names, then this setting will be ignored. AlternativeName= The alternative interface name to use. This option can be specified multiple times. If the empty string is assigned to this option, the list is reset, and all prior assignments have no effect. If the kernel does not support the alternative names, then this setting will be ignored. TransmitQueues= Specifies the device's number of transmit queues. An integer in the range 1…4096. When unset, the kernel's default will be used. ReceiveQueues= Specifies the device's number of receive queues. An integer in the range 1…4096. When unset, the kernel's default will be used. TransmitQueueLength= Specifies the transmit queue length of the device in number of packets. An unsigned integer in the range 0…4294967294. When unset, the kernel's default will be used. MTUBytes= The maximum transmission unit in bytes to set for the device. The usual suffixes K, M, G, are supported and are understood to the base of 1024. BitsPerSecond= The speed to set for the device, the value is rounded down to the nearest Mbps. The usual suffixes K, M, G, are supported and are understood to the base of 1000. Duplex= The duplex mode to set for the device. The accepted values are half and full. AutoNegotiation= Takes a boolean. If set to yes, automatic negotiation of transmission parameters is enabled. Autonegotiation is a procedure by which two connected ethernet devices choose common transmission parameters, such as speed, duplex mode, and flow control. When unset, the kernel's default will be used. Note that if autonegotiation is enabled, speed and duplex settings are read-only. If autonegotiation is disabled, speed and duplex settings are writable if the driver supports multiple link modes. WakeOnLan= The Wake-on-LAN policy to set for the device. The supported values are: phy Wake on PHY activity. unicast Wake on unicast messages. multicast Wake on multicast messages. broadcast Wake on broadcast messages. arp Wake on ARP. magic Wake on receipt of a magic packet. secureon Enable secureon(tm) password for MagicPacket(tm). off Never wake. Defaults to off. Port= The port option is used to select the device port. The supported values are: tp An Ethernet interface using Twisted-Pair cable as the medium. aui Attachment Unit Interface (AUI). Normally used with hubs. bnc An Ethernet interface using BNC connectors and co-axial cable. mii An Ethernet interface using a Media Independent Interface (MII). fibre An Ethernet interface using Optical Fibre as the medium. Advertise= This sets what speeds and duplex modes of operation are advertised for auto-negotiation. This implies AutoNegotiation=yes. The supported values are: Advertise Speed (Mbps) Duplex Mode 10baset-half 10half10baset-full 10full100baset-half 100half100baset-full 100full1000baset-half 1000half1000baset-full 1000full10000baset-full 10000full2500basex-full 2500full1000basekx-full 1000full10000basekx4-full 10000full10000basekr-full 10000full10000baser-fec 10000full20000basemld2-full 20000full20000basekr2-full 20000full By default this is unset, i.e. all possible modes will be advertised. This option may be specified more than once, in which case all specified speeds and modes are advertised. If the empty string is assigned to this option, the list is reset, and all prior assignments have no effect. ReceiveChecksumOffload= Takes a boolean. If set to true, the hardware offload for checksumming of ingress network packets is enabled. When unset, the kernel's default will be used. TransmitChecksumOffload= Takes a boolean. If set to true, the hardware offload for checksumming of egress network packets is enabled. When unset, the kernel's default will be used. TCPSegmentationOffload= Takes a boolean. If set to true, the TCP Segmentation Offload (TSO) is enabled. When unset, the kernel's default will be used. TCP6SegmentationOffload= Takes a boolean. If set to true, the TCP6 Segmentation Offload (tx-tcp6-segmentation) is enabled. When unset, the kernel's default will be used. GenericSegmentationOffload= Takes a boolean. If set to true, the Generic Segmentation Offload (GSO) is enabled. When unset, the kernel's default will be used. GenericReceiveOffload= Takes a boolean. If set to true, the Generic Receive Offload (GRO) is enabled. When unset, the kernel's default will be used. LargeReceiveOffload= Takes a boolean. If set to true, the Large Receive Offload (LRO) is enabled. When unset, the kernel's default will be used. RxChannels= Sets the number of receive channels (a number between 1 and 4294967295) . TxChannels= Sets the number of transmit channels (a number between 1 and 4294967295). OtherChannels= Sets the number of other channels (a number between 1 and 4294967295). CombinedChannels= Sets the number of combined set channels (a number between 1 and 4294967295). RxBufferSize= Takes an integer. Specifies the maximum number of pending packets in the NIC receive buffer. When unset, the kernel's default will be used. RxMiniBufferSize= Takes an integer. Specifies the maximum number of pending packets in the NIC mini receive buffer. When unset, the kernel's default will be used. RxJumboBufferSize= Takes an integer. Specifies the maximum number of pending packets in the NIC jumbo receive buffer. When unset, the kernel's default will be used. TxBufferSize= Takes an integer. Specifies the maximum number of pending packets in the NIC transmit buffer. When unset, the kernel's default will be used. RxFlowControl= Takes a boolean. When set, enables the receive flow control, also known as the ethernet receive PAUSE message (generate and send ethernet PAUSE frames). When unset, the kernel's default will be used. TxFlowControl= Takes a boolean. When set, enables the transmit flow control, also known as the ethernet transmit PAUSE message (respond to received ethernet PAUSE frames). When unset, the kernel's default will be used. AutoNegotiationFlowControl= Takes a boolean. When set, the auto negotiation enables the interface to exchange state advertisements with the connected peer so that the two devices can agree on the ethernet PAUSE configuration. When unset, the kernel's default will be used. GenericSegmentOffloadMaxBytes= Specifies the maximum size of a Generic Segment Offload (GSO) packet the device should accept. The usual suffixes K, M, G, are supported and are understood to the base of 1024. An unsigned integer in the range 1…65536. Defaults to unset. GenericSegmentOffloadMaxSegments= Specifies the maximum number of a Generic Segment Offload (GSO) segments the device should accept. An unsigned integer in the range 1…65535. Defaults to unset. The link file 99-default.link that is shipped with systemd defines the default naming policy for links. [Link] NamePolicy=kernel database onboard slot path MACAddressPolicy=persistent This example assigns the fixed name dmz0 to the interface with the MAC address 00:a0:de:63:7a:e6: [Match] MACAddress=00:a0:de:63:7a:e6 [Link] Name=dmz0 NamePolicy= is not set, so Name= takes effect. We use the 10- prefix to order this file early in the list. Note that it needs to be before 99-link, i.e. it needs a numerical prefix, to have any effect at all. $ sudo SYSTEMD_LOG_LEVEL=debug udevadm test-builtin net_setup_link /sys/class/net/hub0 … Parsed configuration file /usr/lib/systemd/network/99-default.link Parsed configuration file /etc/systemd/network/10-eth0.link ID_NET_DRIVER=cdc_ether Config file /etc/systemd/network/10-eth0.link applies to device hub0 link_config: autonegotiation is unset or enabled, the speed and duplex are not writable. hub0: Device has name_assign_type=4 Using default interface naming scheme 'v240'. hub0: Policies didn't yield a name, using specified Name=hub0. ID_NET_LINK_FILE=/etc/systemd/network/10-eth0.link ID_NET_NAME=hub0 … Explicit Name= configuration wins in this case. sudo SYSTEMD_LOG_LEVEL=debug udevadm test-builtin net_setup_link /sys/class/net/enp0s31f6 … Parsed configuration file /usr/lib/systemd/network/99-default.link Parsed configuration file /etc/systemd/network/10-eth0.link Created link configuration context. ID_NET_DRIVER=e1000e Config file /usr/lib/systemd/network/99-default.link applies to device enp0s31f6 link_config: autonegotiation is unset or enabled, the speed and duplex are not writable. enp0s31f6: Device has name_assign_type=4 Using default interface naming scheme 'v240'. enp0s31f6: Policy *keep*: keeping existing userspace name enp0s31f6: Device has addr_assign_type=0 enp0s31f6: MAC on the device already matches policy *persistent* ID_NET_LINK_FILE=/usr/lib/systemd/network/99-default.link … In this case, the interface was already renamed, so the keep policy specified as the first option in 99-default.link means that the existing name is preserved. If keep was removed, or if were in boot before the renaming has happened, we might get the following instead: enp0s31f6: Policy *path* yields "enp0s31f6". enp0s31f6: Device has addr_assign_type=0 enp0s31f6: MAC on the device already matches policy *persistent* ID_NET_LINK_FILE=/usr/lib/systemd/network/99-default.link ID_NET_NAME=enp0s31f6 … Please note that the details of output are subject to change. This example assigns the fixed name internet0 to the interface with the device path pci-0000:00:1a.0-*: [Match] Path=pci-0000:00:1a.0-* [Link] Name=internet0 Here's an overly complex example that shows the use of a large number of [Match] and [Link] settings. [Match] MACAddress=12:34:56:78:9a:bc Driver=brcmsmac Path=pci-0000:02:00.0-* Type=wlan Virtualization=no Host=my-laptop Architecture=x86-64 [Link] Name=wireless0 MTUBytes=1450 BitsPerSecond=10M WakeOnLan=magic MACAddress=cb:a9:87:65:43:21 systemd-udevd.service8 , udevadm8 , systemd.netdev5 , systemd.network5