Mailing List Archive: Cisco: NSP

Re: OT: VSS + MEC - port-channel dynamically cloned?

Index | Next | Previous | View Threaded

kgraham at industrial-marshmallow

Nov 24, 2009, 11:03 AM

Post #1 of 4 (725 views)
Permalink

Re: OT: VSS + MEC - port-channel dynamically cloned?

[...taking this from nanog to c-nsp...]

> Essentially, for all of the MEC connections, the VSS has created a clone
> of the configured port-channel to bind the actual physical connections,
> rather than binding them under the configured port-channel (and suffixed
> the port-channel number with A or B depending on which chassis was first
> to bind).

I believe this is an LACP artifact; speculation, but when the port-channel
is first formed, the far-end aggregator is saved. If the channel is
re-formed with a new aggregator, the channel is "cloned" like this.

I've only tripped across this on standalone 6500's when bringing up new
LACP bundles; destroying and recreating them worked fine (though as you
noticed, there's no functional impact from a "cloned" Po interface, just
cosmetic).
_______________________________________________
cisco-nsp mailing list cisco-nsp [at] puck
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/

achatz at forthnet

Nov 24, 2009, 12:01 PM

Post #2 of 4 (710 views)
Permalink

Re: OT: VSS + MEC - port-channel dynamically cloned? [In reply to]

I have seen (very frequently) cloned A and B port-channels (debug calls them "secondary
aggregators" if i remember right) created on a 6500 after reloading the peer router
(C10k). Quite annoying, since the cloned interface is a new interface and snmp counters do
not work anymore (neither our eem scripts) .

According to tac, this is expected behavior for LACP if there is a misconfiguration
(typically when two links of the same channel are attempted to be connected on two
different devices on the remote end, like in this case, VSS/MEC) on the moment of the
channel bundling. But in my case it was something else and tac couldn't find the root cause.

--
Tassos

Kevin Graham wrote on 24/11/2009 21:03:
> [...taking this from nanog to c-nsp...]
>
>
>
>> Essentially, for all of the MEC connections, the VSS has created a clone
>> of the configured port-channel to bind the actual physical connections,
>> rather than binding them under the configured port-channel (and suffixed
>> the port-channel number with A or B depending on which chassis was first
>> to bind).
>
> I believe this is an LACP artifact; speculation, but when the port-channel
> is first formed, the far-end aggregator is saved. If the channel is
> re-formed with a new aggregator, the channel is "cloned" like this.
>
> I've only tripped across this on standalone 6500's when bringing up new
> LACP bundles; destroying and recreating them worked fine (though as you
> noticed, there's no functional impact from a "cloned" Po interface, just
> cosmetic).
> _______________________________________________
> cisco-nsp mailing list cisco-nsp [at] puck
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/
>

_______________________________________________
cisco-nsp mailing list cisco-nsp [at] puck
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/

mtinka at globaltransit

Nov 24, 2009, 4:10 PM

Post #3 of 4 (689 views)
Permalink

Re: OT: VSS + MEC - port-channel dynamically cloned? [In reply to]

On Wednesday 25 November 2009 04:01:47 am Tassos
Chatzithomaoglou wrote:

> According to tac, this is expected behavior for LACP if
> there is a misconfiguration (typically when two links of
> the same channel are attempted to be connected on two
> different devices on the remote end, like in this case,
> VSS/MEC) on the moment of the channel bundling. But in
> my case it was something else and tac couldn't find the
> root cause.

We've seen this behaviour when we enabled LACP under SXH3.

However, since upgrading to SXI2a late last month, there are
no more cloned interfaces. The actual port-channel is the
only port-channel. I found the "cloning" weird, at first.
But clearly, as much as it wasn't documented, SXI2a "fixed"
that issue :-).

We're also not seeing this issue on a 7606/RSP720-3CXL
running 12.2(33)SRC5.

Cheers,

Mark.

Attachments:

signature.asc (0.82 KB)

julien.couturier at gmx

Nov 25, 2009, 6:03 AM

Post #4 of 4 (665 views)
Permalink

Re : Re: OT: VSS + MEC - port-channel dynamically cloned? [In reply to]

_______________________________________________
cisco-nsp mailing list cisco-nsp [at] puck
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/

Index | Next | Previous | View Threaded

Interested in having your list archived? Contact Gossamer Threads