[c-nsp] C6k diag failure in lab, need to worry?

[Peter Rathlev]

On Wed, 2008-04-09 at 23:11 -0700, e ninja wrote:
> Sukumar,
> 
> " You can ignore this one, as it _should_ not have any impact, after
> the second reload." is not an acceptable answer. 
> 
> 1 crash in every 100 reboots = 1 million crashes out of every 100
> million reboots. In our quest for perfection, we should strive to
> investigate and rectify every unexpected deviation from the norm. 
> 
> Peter,
> 
> Open a TAC case and submit all the captures for Cisco BU to
> investigate and rectify so that all other customers can benefit from
> the solution.

Well, if this only happens "seldomly", and only during reboots, and
effectively just extends the reboot time a few minutes, I would say that
there are other things I'd like fixed first. :-)

On the other hand, I don't see why we shouldn't at least give TAC a
notice, so they can see a scenario where it happens. So I think we'll
open a low priority case on it.

That made me realise that the "show tech | redirect
disk0:/show-tech.txt" doesn't work quite as I expected. Even with the
redirect, I get a lot of output to the console and the resulting file
doesn't seem complete (just a guess though). I only had console access,
no VTY. I tried just letting a "show tech" run in a terminal window with
logging, but the result was only around 8000 lines; our production boxes
spit out around 16000 lines from a similar command, though with other
cards. But maybe the serial link didn't work equally well all the time.

Isn't there a way to have the box gather the relevant information and
then store in on disk0:, so I can move it from there? I can connect my
laptop to the supervisor copper interface and configure IP from there,
it just seems a little unflexible.

Regards,
Peter