This should be a better default than using VDSL2 since "conservative"
should work on all connections.
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
Signed-off-by: Arne Fitzenreiter <arne_f@ipfire.org>
Some devices do not have an address, which prints an error when reading
it.
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
Signed-off-by: Arne Fitzenreiter <arne_f@ipfire.org>
This is a more generic term since CoDel is no longer being used
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
Signed-off-by: Arne Fitzenreiter <arne_f@ipfire.org>
This script configures CAKE as default qdisc where appropriate:
* It is not suitable to use any queue management on bridges/GRE/VTI/tun
interfaces.
* On Internet-facing interfaces, CAKE is configured in "internet" mode
whereas on any zones except RED, it is configured in "metro" mode
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
Signed-off-by: Arne Fitzenreiter <arne_f@ipfire.org>
We are using CPU-affinity and packet steering functions in various
places in IPFire, but packets might still be received on a random CPU
core.
This feature enables that packets that belong to the same connection
(i.e. have the save tuple) will be steered to the same queue. This will
increase cache locality and decrease locking which results in higher
throughput.
https://www.kernel.org/doc/Documentation/networking/scaling.txt
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
Acked-by: Peter Müller <peter.mueller@ipfire.org>
Signed-off-by: Arne Fitzenreiter <arne_f@ipfire.org>
This has been broken because of other changes on the network scripts and
since we now have support for bridges there is no point in supporting
something else that has the same functionality.
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
This patch changes that instead of only the wireless interface that is
used to run a wireless access point, all wireless interfaces will be set
to master mode.
This allows that attaching the interface won't fail if hostapd isn't
started, yet.
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
Since using random interface names is always trouble, we will order all
slave interfaces and number them in order as they appear in the SLAVES
variable.
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
Wireless interfaces cannot be added to the bridge when they are
still running in station mode. At boot time, the interface will
be created and it is attempted to attach it to a bridge, which
fails.
This patch now checks whether the wireless interface is going to
be an access point, changes the mode accordingly and then attaches
the interface to the bridge.
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
Tested-by: Daniel Weismüller <daniel.weismueller@ipfire.org>
Signed-off-by: Arne Fitzenreiter <arne_f@ipfire.org>
This allows us to create VLAN interfaces even when the
name of the parent interface might vary.
This patch also appends the VLAN tag to interfaces
when the zone is in bridge mode.
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
The script relied on the configuration being in lowercase.
If people manually editied their configuration file they might
not have paid attention to this and therefore this script now
also accepts uppercase MAC addresses.
Fixes: #12047
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
Instead of orange0phys we should use orangephys0 this patch implements
the necessary changes.
Signed-off-by: Jonatan Schlag <jonatan.schlag@ipfire.org>
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
It is now also possible to use the MAC address to define a slave of a
bridge.
Simply add the mac address to the ZONE_SLAVES=''.
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
This script is creating common bridges now, too and therefore
needs a more generic name.
Signed-off-by: Jonatan Schlag <jonatan.schlag@ipfire.org>
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
This bridge mode is supposed to be used for virtual environments
to create a network zone as a bridge and have virtual machines inside
it. Other physical interfaces can also be added to the bridge.
This is very similar to the MACVTAP bridge feature but still works
when the link of any (or all) physical interfaces is down.
Fixes: #11252
Signed-off-by: Jonatan Schlag <jonatan.schlag@ipfire.org>
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
This change make it possible to use a macvtap interface as a
standard interface (green0).
This is required by libvirt, because libvirt adds macvtap interfaces to
the physical interface, but this causes a problem. A VM with this
configuration can communicate with the whole network,
but not with the Host (IPFire).
To solve this problem, the host interface must be also a macvtap interface.
This is achieved by:
1. In /var/ipfire/ethernet/settings the mode of a interface could set
with GREEN_MODE= ...
When the mode is macvtap the physical interface is renamed to green0phys
instead of green0. If the mode is not set the normal configuration is
applied .
2. The network-hotplug-macvtap script checks if a physical nic ends
with "phys".
When the interface ends with "phys", the script adds a macvtap interface
to the physical nic which is named green0. The MAC address of this
interface is set to the MAC address of the physical nic. The MAC address
of the physical is set to a random value. We do this because the MAC
address of green0 should not change.
All services, IP addresses then binds to the macvatap interface, the
physical nic is not used.
PS.: The script works also with the orange or blue interface, just
replace green with orange or blue.
Signed-off-by: Jonatan Schlag <jonatan.schlag@ipfire.org>
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
The VLAN hotplugging script was called with a wrong parameter that
should just be used to rename devices. Hence the script was not
correctly executed and did not create the virtual devices.
Fixes: #11009
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
The VLAN devices will now automatically be created after
a parent device has been added.
Mainly this will resolve a race-condition between udev
initialising the network adapters and sysvinit running
scripts that will do the initialisation of the VLAN.
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
Instead of creating a copy of the configuration values and
for better extensibility, we will have udev execute a script
that parses /var/ipfire/ethernet/settings and will return the
correct name of the corresponding device (green0, blue0, ...).
This will change the format of the mISDN config file
back to plain config not XML.
For the new old script there is bc required.
Additionally the mISDN modules and some related are not
loaded by udev on sys-start but manually by the
mISDN init script.
