Bug#878162: systemd-networkd ignores MTU from RA
cyril at debamax.com
Tue Oct 10 16:53:44 BST 2017
Tags: patch ipv6
A customer of mine reported a regression by switching from jessie to
stretch: /etc/systemd/network/eth.network is configured to only set up
IPv4, and systemd in jessie used to do only that, and not to touch IPv6
Anyway: accept_ra_mtu defaults to true, is set to true on the target
system, but systemd doesn't honour MTU settings, and defaults to the
interface's MTU for the IPv6 routes it sets up.
This is fixed upstream between 232 and 233:
| commit d6fceaf1f7ff765bdc3b135f3d3676ec689da312
| Author: Susant Sahani <ssahani at users.noreply.github.com>
| Date: Thu Nov 24 03:02:19 2016 +0530
| networkd: handle MTU field in IPv6 RA (#4719)
| This patch handles the custom MTU field in IPv6 RA.
| fixes RFE #4464
I haven't checked that the Debian packages in buster/unstable indeed
avoid running into this issue, but I've confirmed that cherry-picking
this patch on top of the stretch package fixes this issue. I'm attaching
the source debdiff which was successfully tested.
Thanks for considering a fix through stretch-proposed-updates.
Cyril Brulebois -- Debian Consultant @ DEBAMAX -- https://debamax.com/
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 7298 bytes
Desc: not available
More information about the Pkg-systemd-maintainers