Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 11 Feb 2020 10:36:43 -0500
From:      Mark Saad <nonesuch@longcount.org>
To:        Eugene Grosbein <eugen@grosbein.net>
Cc:        FreeBSD Hackers <freebsd-hackers@freebsd.org>
Subject:   Re: LACP / Slow-protocol issue
Message-ID:  <CAMXt9NY3sbTWo-CV=gGf=UUnABJET%2B5TfOhSe=C_euGEo8MWvQ@mail.gmail.com>
In-Reply-To: <bb566a20-7fbb-03f3-cb6a-83371e39486d@grosbein.net>
References:  <CAMXt9NYVyk7xpevMJE65eG8D2JPg2aY3X2Sh4r2WusqcQMzkGA@mail.gmail.com> <bb566a20-7fbb-03f3-cb6a-83371e39486d@grosbein.net>

next in thread | previous in thread | raw e-mail | index | archive | help
Eugene
 On the upstream switches; I have span of the interfaces going into a
packet capture device .  I can pull pcaps
going back a few weeks, however what I see  for ethertype 0x8809 is
not all that interesting; and I am not exactly sure of the
intended interaction of switch to lagg  with lacp.  Also, I seriously
doubt this is related to iflib. The only reason I say this is, some of
my devices with the same-ish* setup do not exhibit the issue.

*some of the switches are running on newer vendor hardware.

On Tue, Feb 11, 2020 at 1:43 AM Eugene Grosbein <eugen@grosbein.net> wrote:
>
> 11.02.2020 2:30, Mark Saad wrote:
>
> > Hello FreeBSD Hackers
> >   I was wondering if there is anyone on-list who has hands on
> > knowledge of the "slow-protocols" and in particular lacp .
> >
> > I am running a fairly new 12.1-STABLE amd64 r354698 , for reasons I
> > can't yet identify a two port lacp lagg has a short <1sec blip where
> > switches and lagg members believe each side has stopped distributing
> > lacp slow-protocol frames.  I am not super knowledgeable about how to
> > diagnose the slow-protocol frames to see who is at fault . Any insight
> > here would help.
>
> Just use: tcpdump -s0 -np -i parent0 not ip
> Replace "parent0" with name of your real (physical) interface name (one of), like igb0.
>
> You may be facing one of iflib's problems, so try booting FreeBSD 11.3-RELEASE (pre-iflib)
> to compare behaviour and see if it runs better or not.
>
>


-- 
mark saad | nonesuch@longcount.org



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAMXt9NY3sbTWo-CV=gGf=UUnABJET%2B5TfOhSe=C_euGEo8MWvQ>