Age | Commit message (Collapse) | Author |
|
Set the MRU (Maximum Receive Unit) value to n. PPPd will ask the peer to send
packets of no more than n bytes. The value of n must be between 128 and 16384,
the default was always 1492 to match PPPoE MTU.
A value of 296 works well on very slow links (40 bytes for TCP/IP header + 256
bytes of data). Note that for the IPv6 protocol, the MRU must be at least 1280.
CLI:
set interfaces pppoe pppoe0 mru 1280
(cherry picked from commit e062a8c11856f213983f5b41f50d4f9dbc0dde0f)
|
|
Replace links to the phabricator site from https://phabricator.vyos.net to
https://vyos.dev
(cherry-picked form commit bd9416a6aa9d5d0a746dc2cebc8d0330fd27d1a2)
|
|
Some ISPs seem to use the host-uniq flag to authenticate client equipment.
Add CLI option in VyOS to allow specification of the host-uniq flag.
set interfaces pppoe pppoeN host-uniq <value>
(cherry-picked from commit 38bab79324087df5a9057c23b85a0a784c09540a)
|
|
This is a backport of https://github.com/vyos/vyos-1x/pull/1656.
Note I also changed `ip-down.script.tmpl` to not wait for `systemctl
stop dhcp6c@$iface.service`, because that command is slow and pppd will
kill the ip-down script if it times out.
I didn't see `ip-down.script.tmpl` or its equivalent in the 1.4 branch.
Not sure if there is another mechanism to handle that functionality or
it is missed.
|
|
Adds a sysctl parameter to ignore the default router obtained from
router advertisements when pppoe default-route is set to 'none'.
|
|
(cherry picked from commit 8fc06b5f8bbfcc49e69406fd70cd5cd42fb6d39f)
|
|
(cherry picked from commit b121ee14ff1961b56568b0116de3c246ea4af934)
|
|
(cherry picked from commit 2dc11253e46b945adb392c5946fd9ad43f391d86)
|
|
(cherry picked from commit b64b45c6a5b66b8d6b07ab5a03fccaeabb3677a9)
|
|
|
|
IPv6 enable can be considered once the ipv6 node is present!
|
|
Commit a2ac9fac ("vyos.template: T2720: always enable Jinja2 trim_blocks
feature") globally enabled the trim_blocks feature. Some templates still used
in-line trim_blocks "{%"- or "-%}" which caused miss-placed line endings.
This is fixed by removing all in-line trim_block statememnts of Jinja2 templates.
|
|
DHCPv6 client was not started as it used the old dict keys.
|
|
|
|
Commit 65fa21f5e ("ifconfig: T2653: move pppoe interface to get_config_dict()")
directly takes the configuration keys from the CLI config and thus the keys
used for template generation have not been updated appropriately.
|
|
Commit 03fb97 (pppoe: ipv6: T2681: script bugfix after get_config_dict() migration )
After the PPPoE link is reset, dhcp6c cannot be restarted,
which may cause the prefix delegation of IPv6 to fail to restart.
This submission will restart dhcp6c after the IPv6 of PPPoE is up again
|
|
Commit 65fa21f5 ("ifconfig: T2653: move pppoe interface to get_config_dict()")
did not rename the Jinja2 template vars as required as they are not retrieved
directly as dict from the config backend.
|
|
|
|
|
|
|
|
Commit 8dcf042a7f5 ("pppoe: T2219: move adding of default route to ip-(up|down)
script") wrongly tried to open a file called autoconfigure which infact
should be autoconf
|
|
|
|
Rename the CLI nodes for prefix delegation from "dhcpv6-options delegate
<interface>" to "dhcpv6-options prefix-delegation interface <interface>".
The change is required to add the possibility to request for specific prefix
sized via the CLI. That option was not possible with the old configuration
tree.
|
|
|
|
|
|
|
|
As PPP can be used to establish a connection on-demand it manages the Kernel
default route. This can not be used when using VRFs which are managed by
the ip-up.d and ip-down.d scripts - thus those options are now mutially
exclusive.
The best fix would be adding support for VRFs into PPP.
|
|
|
|
|
|
As PPPoE interfaces should be part of a VRF it is required that the default
can also be set into the specified VRF. The default way of PPP by setting the
"defaultroute" option does not take a VRF into account. In this case PPP
installs a Kernel route forwarding all default traffic over this interface.
This change installs a static route with metric 1 instead into the routing
table via FRR. This is not the same as a Kernel route in terms of metric 1
compared to 0 but it should do the trick.
|
|
This makes the actual code which generates the configs much more human
readable.
|