Date: Mon, 13 Jul 2015 09:56:31 -0700 From: John-Mark Gurney <jmg@funkthat.com> To: Karl Pielorz <kpielorz_lst@tdx.co.uk> Cc: freebsd-net@freebsd.org Subject: Re: lagg of em0/em1 + VLAN = lower MTU? Message-ID: <20150713165631.GN8523@funkthat.com> In-Reply-To: <EF441569FCACD78034320411@[10.12.30.106]> References: <7CFE75F7566F5789DAD9FBB2@[10.12.30.106]> <20150710180627.GA8523@funkthat.com> <EF441569FCACD78034320411@[10.12.30.106]>
next in thread | previous in thread | raw e-mail | index | archive | help
Karl Pielorz wrote this message on Mon, Jul 13, 2015 at 09:33 +0100:
> --On 10 July 2015 11:06 -0700 John-Mark Gurney <jmg@funkthat.com> wrote:
>
> > Try bumping the MTU on the root em's by 4 (1504) before creating the
> > lagg...
>
> I had thought of that - but didn't want to try it (on the basis that out of
> all the other example config's I've seen - no one else has to, and it will
> break the lagg MTU).
>
> Anyway I tried it, by:
>
> - Bump the MTU on the em interfaces before creating the lagg to 1504
>
> - The lagg created, inherits this 1504 MTU
>
> That breaks the untagged IP on lagg0 - i.e. I end up with:
>
> "
> lagg0: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> metric 0 mtu 1504
>
> options=4019b<RXCSUM,TXCSUM,VLAN_MTU,VLAN_HWTAGGING,VLAN_HWCSUM,TSO4,VLAN_HWTSO>
> ether 68:05:ca:08:1d:3b
> inet x.x.x.x netmask 0xffffff00 broadcast y.y.y.y
> "
>
> That MTU of 1504 is going to break lagg0 for untagged traffic :(
>
>
> - However, the VLAN's setup on the lagg still default to 1496 MTU (despite
> the underlying interface lagg0 being MTU 1504)
>
> If I 'manually' bump the MTU on the individual lagg0.x VLAN interfaces -
> they will go to 1500, but traffic is still 'broken'.
>
> I can't 'reduce' the bad MTU on lagg0 either - i.e. 'ifconfig lagg0 mtu
> 1500' (to try to put it back to 1500) fails with an ioctl error.
>
> So sadly, I'm still stuck :(
You can fix the untagged traffic on lagg0 with something like the
following:
route change <destination> <gateway> -mtu 1500
You can see which routes need to be changed w/ the netstat -rWnfinet
command...
> Something is lopping 4 bytes off the MTU when it shouldn't need to. None of
> the example configs I could find were from 10.1-R (or with exactly the same
> cards) - but obviously this is just bread & butter lagg/VLAN stuff, so it
> should work?
You should be able to make it work, but it definately needs more work
to make it work automaticly... There was a recent change that went
in that makes changing lagg mtu easier, but I think that's only in
HEAD right now...
Hope this helps...
--
John-Mark Gurney Voice: +1 415 225 5579
"All that I will do, has been done, All that I have, has not."
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20150713165631.GN8523>