commit 76ea485d9edb781328e307c68b1f878d933408e5
Author: Michael Tremer <michael.tremer@ipfire.org>
Date: Fri Sep 27 17:39:22 2024 +0200
wireguard: Select the correct source IP address for N2N peers
This is so that the firewall chooses the correct IP address when trying
to establish connections to the remote networks.
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
Signed-off-by: Vincent Li <vincent.mc.li@gmail.com>
This option needs to be configurable since some (braindead) ISPs have
started running broken DHCP servers to be bug-compatible with cheap
broken plastic routers.
By default we keep this option enabled, but it can now be turned off
whenever needed.
Suggested-by: Adolf Belka <adolf.belka@ipfire.org>
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
Tested-by: Adolf Belka <adolf.belka@ipfire.org>
Signed-off-by: Arne Fitzenreiter <arne_f@ipfire.org>
- This ensures that all ip route and ip rule commands are redirected to null if the output
is not used to feed into a variable.
- This will prevent any error messages related to empty iproute tables being displayed
during boot if an empty table is accessed.
Fixes: Bug#12763
Tested-by: Adolf Belka <adolf.belka@ipfire.org>
Signed-off-by: Adolf Belka <adolf.belka@ipfire.org>
Signed-off-by: Arne Fitzenreiter <arne_f@ipfire.org>
This updated version of this script avoids any errors if collectd is not
running (yet) which might happen during the boot process.
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
The "ping" plugin does not re-resolve the gateway IP address after
pinging it for the first time. For most people this won't be a big
problem, but if the default gateway changes, the latency graph won't
work any more.
In order to do re-resolve "gateway", the only way is to restart
collectd.
Fixes: #13522
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
Acked-by: Peter Müller <peter.mueller@ipfire.org>
There is no sense in doing this only in /etc/init.d/network
and not in /etc/init.d/networking/red
This files should be always deleted before a startup
Signed-off-by: Jonatan Schlag <jonatan.schlag@ipfire.org>
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
/bin/sh is a symlink to /bin/bash on ipfire systems. Using /bin/sh in
the scripts as shebang hurts in two ways:
1. We use features which do not work with sh as shell. This is not
really a problem but if we rely on features of a real bash we can
state this clearly.
2. The syntay highlighting in vim does not work without a correct
shebang. As I want and need correct syntax highlighting I propose to
change the shebang.
Signed-off-by: Jonatan Schlag <jonatan.schlag@ipfire.org>
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
- This patch goes together with the patch for the ppp update to 2.5.0
- The rp-pppoe.so option is no longer available. There is only the pppoe.so available now
Fixes: Bug#13164
Signed-off-by: Adolf Belka <adolf.belka@ipfire.org>
Reviewed-by: Michael Tremer <michael.tremer@ipfire.org>
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
This is really badly hacky, but I do not know a better way to solve this
with our existing "setup" program which would be a nightmare to extend.
So we are using the device number to generate a static MAC address which
can then be used as usual. I doubt many people will have more than one
device.
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
When acquiring an IP address, dhcpcd seems to think that the interface
is down or does not work properly for some reason. It will
subsequentially decide to exit which is not what we want here.
Therefore this patch tells dhcpcd to ignore the link state and keep
happily running.
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
This could potentially create problems when we abuse these functions to
launch the DHCP client on IPTV interfaces. This would have to be tested
and confirmed or potentially we would need some more changes to keep
supporting that use-case, too.
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
QMI is a proprietary interface from Qualcomm which are absolute pioneers
when it comes to interfacing with modems. I don't think there would be
any way to make this any more complicated and bloated.
So, bascially we will put the modem into a raw IP mode which changes the
interface into Point-to-Point mode.
We then configure the provider settings using qmicli. After that, the
modem will try to connect to the provider and obtain an IP address.
We will then start a DHCP client which does not do any DHCP-ing because
implementing that would be too complicated. Instead we do something even
*more* complicated where we would launch a custom script which asks the
modem for the allocated IP address and will configure it into the
device. The DHCP client then reads that IP address from the device and
pretends it came up with it by itself. Such an easy way to do this.
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
Fixes: 12831
Jonatan Schlag reported that the command line options of 'vnstat' had changed
"...and seemed to be broken a long time".
=> https://bugzilla.ipfire.org/show_bug.cgi?id=12831#c0
Several command line switches used in networking initscripts were obviously removed.
Affected commands in '.../networking/any' and '.../networking/red'):
...
/usr/bin/vnstat -u -i ${DEVICE} -r --enable --force > /dev/null 2>&1
...
/usr/bin/vnstat -u -i ${DEVICE} -r --disable > /dev/null 2>&1
...
and
...
/usr/bin/vnstat -u -i ppp0 -r --disable > /dev/null 2>&1
...
Adolf Belka tested this, "looked through the changelogs" and found - besides that
the switch '--enable' had been removed "in version 2.0 in 2018" - that '--enable', '--update'
and '--reset' switches are either not needed or not supported anymore.
"The old man page indicates that none of those options are used when the vnstat daemon
is running."
Since we only start and run 'vnstatd' in IPFire it was decided to remove these commands.
Reported-by: jonatan.schlag <jonatan.schlag@ipfire.org>
Tested-by: Adolf Belka <adolf.belka@ipfire.org>
Signed-off-by: Matthias Fischer <matthias.fischer@ipfire.org>
The DHCP server can instruct clients to configure a certain MTU.
This used to be done by setting the MTU of the interface. However,
dhcpcd has changed this behaviour using routes to.
We used to have a modified version of the old mechanism which no longer
works well with the new system and is therefore to be dropped.
This is the first commit in the series implementing the new behaviour
and telling dhcpcd to use the configured MTU.
Fixes: #12563
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
Reviewed-by: Peter Müller <peter.mueller@ipfire.org>
Bumping across one of our scripts with very long trailing whitespaces, I
thought it might be a good idea to clean these up. Doing so, some
missing or inconsistent licence headers were fixed.
There is no need in shipping all these files en bloc, as their
functionality won't change.
Signed-off-by: Peter Müller <peter.mueller@ipfire.org>
pppd 2.4.9 supports IPv6 and asks for an IPv6 configuration by default.
Setting the received prefix in the kernel will never work, however, as
the rest of IPFire 2.x does not support IPv6.
pppd notices the ISP about this, and at least Otenet (GR) and British
Telecom (several countries) decide to close a dial-up connection then.
German DTAG seems to ignore such errors silently.
This patch adds an option to the pppd call to prevent asking for an
IPv6 configuration, hence avoiding this errors.
To apply this patch, it is necessary to ship ppp 2.4.9 again. Since I
have no access to a testing machine behind an ISP supporting IPv6, this
patch unfortunately is untested.
Fixes: #12651
Signed-off-by: Peter Müller <peter.mueller@ipfire.org>
Signed-off-by: Arne Fitzenreiter <arne_f@ipfire.org>
with kernel 5.10 dhcpcd hung at shutdown if red was a wireless client
becuase there was two running instances. This change repeat the
dcpcd -k call.
Signed-off-by: Arne Fitzenreiter <arne_f@ipfire.org>
This variable is no longer being used and was only used to
assign IP addresses to the individual interfaces.
However, the kernel knows best which IP address to select
as broadcast address for each network. Therefore we depend
on the kernel which allows us to support RFC3021.
Fixes: #12486 - no /31 transfer net available on red
Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>
