Hello,

My campus runs a PacketFence 4.1.0 server on Debian 7 that has all
student traffic inline with a 1 Gbps link from the core router and a 1
Gbps link back to the core router. This creates a 1 Gbps bottleneck,
which wasn't a problem until we upgraded our campus bandwidth to 2
Gbps.

I tried to solve this using interface bonding by using the four
available NIC ports to make two 2 Gbps LACP links. I made some
excellent progress making modifications to my /etc/network/interfaces
file with bonding directives and some changes to the [interfaces
intX.VLAN#] directives in pf.conf. Things looked very promising, as I
could SSH to the management IP address on a configured bond interface.

However, when I rebooted the server, the /etc/network/interfaces file
got overwritten by PacketFence. Is there any supported way in pf.conf
or other config files to cause the /etc/network/interfaces config file
to be written the way I want? The directives I am interested in having
written to the config file are 'bond-mode', 'bond-miimon',
'bond-lacp-rate', and 'slaves'.

The slave interfaces had lines in /etc/network/interfaces like 'iface
eth0 inet manual' that got overwritten to 'iface eth0 inet static'. Is
there a way to ensure those lines get (over)written as 'manual' and
not 'static'?

I have thought about creating a startup script that overwrites the
/etc/network/interfaces file on boot to what I want, then restart
networking. However, that is far from ideal, as any restart of
PacketFence or changing settings in the web GUI could leave the config
in a really unstable state until another reboot.

Any ideas are greatly appreciated.

Thanks,
Michael M


_______________________________________________
PacketFence-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/packetfence-users
  • [PacketFence-users] Link bond... Michael M via PacketFence-users

Reply via email to