[c-nsp] 3640 with misbehaving NVRAM

[Bill Wichers]

[snip]
> 3620, no go....even went as far as to yank the nvram, boot, jam it back
> in and try to sync, no go).

That may have failed the chip. They're not hot-swap. When power is not
brought up in the right order (ground -> power -> data pins), sometimes
the chips will power up through clamp diodes or gates which can cause
internal failures in the chips.

> This very well may be as simple as failed nvram (any idea where I can
> get one of those for equipment that is no longer on contract?.....this
> is just lab equipment), but based on some of the things I found, that
> may not be it...and replacing the nvram with a new one may only work
> until the next powerfail.
>
> Anyone deal with this one before and have some insight?

I haven't ever seen the NVRAM get corrupt except once on a bad router (I
think a 1600), and we always attributed that to some kind of
lightning-induced problem on a T1 circuit. You could try looking on Ebay
for NVRAM for the router, or maybe just a whole new router for a low
price. I'd also suggest trying some of the memory suppliers like Crucial
and see if they can help. *Call* them, don't email -- the techs can do
more than the website RAM selector can. We've also gotten some Cisco
memory from http://www.memory.com. The selection of available memory for
Cisco equipment seems limited though, which I think is due to Cisco
occasionally using propreitary memory arrangements in their products.

     -Bill

*****************************
Waveform Technology
UNIX Systems Administrator