[c-nsp] Power Problems - 6509

[Justin M. Streiner]

I should have added "...slots are energized in numerical order when you 
reboot or power-cycle the whole switch..."  When you add a new blade to a 
running chassis, X watts will be requested to power the blade up.  If 
there isn't enough capacity to do so, the switch will complain in the 
logs.

As far as the other issue goes, it does look like it would be a good idea 
to look at that Optera.  You may also want to do some interface-level 
debugging on both of the interfaces that stall out.  Also, do those 
interfaces accumulate any errors when the packets stop flowing?  You 
mentioned replacing jumpers and GBICs - did you do that on both sides of 
the link?

If you replace all of the hardware that can be replaced, the next step 
would be to break out and OTDR and shoot the relevant fiber segments to 
check for problems on the glass itself.  That, plus some knowledge of the 
fiber you're running on will usually flush out physical plant problems, 
i.e. "I know this segment is 3,500 feet long, but I'm getting a really 
ugly reflection with lots of loss at 2,500 feet..." and so forth.

jms

On Mon, 27 Nov 2006, Paul Stewart wrote:

> Thanks :)  Our firewall blade in this chassis isn't in production yet ..
> just did a shutdown successfully on it which didn't free up power but
> going to yank it and see.. should free up several amps...
>
> Paul
>
>
> -----Original Message-----
> From: cisco-nsp-bounces at puck.nether.net
> [mailto:cisco-nsp-bounces at puck.nether.net] On Behalf Of Justin M.
> Streiner
> Sent: Monday, November 27, 2006 9:23 AM
> To: cisco-nsp at puck.nether.net
> Subject: Re: [c-nsp] Power Problems - 6509
>
> On Mon, 27 Nov 2006, Paul Stewart wrote:
>
>> The mode is redundant but if the power gets much lower, will the power
>
>> "split" between the power supplies giving up redundancy or what will
>> the outcome be?
>
> It _might_ split for now, but be prepared for the possibility of strange
> things happening, like modules shutting down on their own.  You may also
> get warnings that the chassis can't supply all of the requested power,
> so some modules will not be energized.  The slots are normally energized
> in numerical order, so whatever modules you have at the end will not get
> powered up if you're short on juice.
>
> If/when you upgrade to Sup720s, I believe those slots will get priority
> on power since nothing else will come up if the sups can't be energized.
>
> jms
>