[c-nsp] What happens during the "shutdown" and "no shutdown" to a 1000BASE-LX10 port

[Pete Lumbis]

+1

The main thing is that the interface has ties to a number of different
corners of the code (i.e., spanning-tree, vlan interface state,
routing protocols, etc). When you "shut" the interface we send
notifications to all of those interested parties to destroy the data
structures/de-provision the service/turn off.

When we "no shut" we bring the interface back on and tell all of these
services to start over. You hit a very large number of places in the
code with this action.

If you've ever been on a TAC call where they spend an hour or two
trying to find out what's broken only to fix it with an interface flap
it's because they are trying to figure out exactly what component is
misbehaving. They don't want to flap the interface immediately because
that will reset whatever bad state the code got in and all information
about precise root cause will be lost.


-Pete

On Thu, Aug 11, 2011 at 11:53 AM, Mack McBride <mack.mcbride at viawest.com> wrote:
> Shut/no shut can fix a multitude of transient faults.
>
> Doing analysis after the fact is extremely difficult.
>
> Mack
>
> -----Original Message-----
> From: cisco-nsp-bounces at puck.nether.net [mailto:cisco-nsp-bounces at puck.nether.net] On Behalf Of Peter Rathlev
> Sent: Thursday, August 11, 2011 4:50 AM
> To: Martin T
> Cc: cisco-nsp at puck.nether.net
> Subject: Re: [c-nsp] What happens during the "shutdown" and "no shutdown" to a 1000BASE-LX10 port
>
> On Thu, 2011-08-11 at 00:56 +0300, Martin T wrote:
>> Gi0/24   -> Brocade  connected   trunk    a-full a-1000 1000BaseLX SFP
>
> When it says "connected" it's not error-disabling as others have tried
> pointing out.
>
>> I'm just curious, what might happened and how did the "shutdown"/"no
>> shutdown" improve the situation? Or is it impossible to analyse such
>> problems(afterwards)?
>
> I'm not sure how Brocade does, but if it was a Cisco switch in the other
> end it could be that e.g. STP loop guard had blocked the link. A link
> down/up event would make STP recalculate and thus open the link again.
>
> Another possibility might be a unidirectional link, though I can't see
> how forcing down/up would resolve that.
>
> I think the problem is easiest to determine with access to the Brocade
> switch. Do the people managing that switch not have logs of some kind to
> consult?
>
> --
> Peter
>
>
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net
> 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.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/
>