clearlinux/clear-linux-documentation
2
0
Fork 0
mirror of https://github.com/clearlinux/clear-linux-documentation.git synced 2026-07-12 07:36:02 +00:00
Code Issues Packages Projects Releases Wiki Activity

Merge pull request #5 from dbkinder/network-bonding

Browse Source 
network: techwriter pass on network bonding guide
This commit is contained in:
mltullis
2017-06-05 13:16:01 -07:00
committed by GitHub
parent 4ac7aa7761 0e212e9e86
commit b3a7de4ca1
1 changed files with 57 additions and 71 deletions
Download Patch File Download Diff File Expand all files Collapse all files
source/clear-linux/network/network-bonding.rst
+57 -71
View File
@@ -4,118 +4,104 @@ Network Bonding
###############
Network bonding is a technique for combining multiple network interfaces into
a single, logical interface, providing some combination of redundancy and
bandwidth aggregation.
a single, logical interface, providing some redundancy and bandwidth
aggregation.
|CLOSIA| includes the bonding_ and team_ drivers. Here, we'll provide an
example of configuring systemd to use the ``bonding`` driver.
|CLOSIA| includes the bonding_ and team_ drivers. The guide example provided
below shows how to configure systemd to use the ``bonding`` driver.
All commands in this guide should be run as root.
.. note::
All commands in this guide must be run as root.
First, if it does not already exist, create the following directory:
1. Create the ``/etc/systemd/network`` directory (if it doesn't already exist):
.. code-block:: console
.. code-block:: console
# mkdir -p /etc/systemd/network
# mkdir -p /etc/systemd/network
In this directory, you will create the configuration files for the virtual
device and the network settings for it and the underlying physical interfaces.
This directory contains the configuration files and network settings
for the virtual device and its underlying physical interfaces.
Next, configure systemd to create a virtual network device, ``bond1``. The
syntax for this file is defined in the systemd.netdev_ manpage.
`This example`__ may be used verbatim, or tuned to your particular requirements.
Note that ``802.3ad`` mode requires explicit support from your NICs and network
switch. This and other modes may also require additional configuration of your
network switch.
2. Configure systemd to create a virtual network device, ``bond1``. Use a text
editor to create a file named ``30-bond1.netdev`` as shown here:
.. code-block:: ini
[NetDev]
Name=bond1
Kind=bond
[Bond]
Mode=802.3ad
TransmitHashPolicy=layer3+4
MIIMonitorSec=1s
LACPTransmitRate=fast
The syntax for this file is defined in the systemd.netdev_ manpage.
`This example`__ may be used verbatim, or tuned to your particular
requirements. Note that ``802.3ad`` mode requires explicit support from
your NICs and network switch. This and other modes may also require
additional configuration of your network switch.
__ https://www.freedesktop.org/software/systemd/man/systemd.netdev.html#id-1.20.10
.. code-block:: ini
:caption: 30-bond1.netdev
:linenos:
3. Configure the slave interfaces, assigning them to the new ``bond1`` device,
using the syntax in systemd.network_, and in a text file named
``30-bond1-enp1s0.network`` as shown here:
[NetDev]
Name=bond1
Kind=bond
.. code-block:: ini
[Bond]
Mode=802.3ad
TransmitHashPolicy=layer3+4
MIIMonitorSec=1s
LACPTransmitRate=fast
[Match]
Name=enp1s0f*
Configure the slave interfaces, assigning them to the new ``bond1``, using the
syntax in systemd.network_.
[Network]
Bond=bond1
.. code-block:: ini
:caption: 30-bond1-enp1s0.network
:linenos:
[Link]
MTUBytes=9000
[Match]
Name=enp1s0f*
[Network]
Bond=bond1
[Link]
MTUBytes=9000
.. note::
This guide demonstrates bonding all four ports of a quad-port NIC, with
This example demonstrates bonding all four ports of a quad-port NIC, with
names in the range ``enp1s0f0-enp1s0f3``, allowing the use of a single file
with a wildcard match. You may also create a separate file for each NIC,
particularly if they have names that are not wildcard-friendly. This
configuration assigns each NIC as a slave of ``bond1``. For best results,
do not assign addresses or DHCP support to the individual NICs.
.. note::
This example also enables jumbo frames of up to 9000 bytes to optimize large
data transfers on the local network. Again, your NICs and switch must
support jumbo frames, and your switch may require additional configuration.
Finally, define the network configuration for the bonded interface.
4. Define the network configuration for the bonded interface in a file named
``30-bond1.network`` as shown here:
.. code-block:: ini
:caption: 30-bond1.network
:linenos:
.. code-block:: ini
[Match]
Name=bond1
[Match]
Name=bond1
[Network]
BindCarrier=enp1s0f0 enp1s0f1 enp1s0f2 enp1s0f3
Address=192.168.1.201/24
[Network]
BindCarrier=enp1s0f0 enp1s0f1 enp1s0f2 enp1s0f3
Address=192.168.1.201/24
[Link]
MTUBytes=9000
.. note::
[Link]
MTUBytes=9000
Since ``bond1`` is a virtual interface, it has no concept of physical link
status. The ``BindCarrier`` directive indicates that the link status of this
interface is determined by the status of the listed slave devices.
.. note::
This is the logical interface, so assign it an IP address. DHCP is more
complicated with bonded interfaces, and is not covered in this guide.
.. note::
complicated with bonded interfaces, and is not covered in this example.
This file also enables jumbo frames of up to 9000 bytes. This option must be
enabled for all slave interfaces *and* the bonded interface in order to take
enabled for all slave interfaces *and* the bonded interface, in order to take
effect.
Apply the new network configuration:
5. Apply the new network configuration:
.. code-block:: console
.. code-block:: console
# systemctl restart systemd-networkd
.. note::
# systemctl restart systemd-networkd
The MTU settings will not take effect until a reboot, or if you explicitly
apply them via ``ifconfig``, for example.