[c-nsp] RAM thing

[Lee]

On 2/17/14, Saku Ytti <saku at ytti.fi> wrote:
> On (2014-02-17 12:24 +0000), Phil Mayers wrote:
>
>> So nothing has changed except we know about it. For anyone who
>> assumed devices could fail at any time, this isn't *that* worrying.
>> For anyone who assumed devices would run forever, this should be a
>> wake-up call - it was never true, and will likely never be so ;o)
>
> Should we expect devices to be build so that broken memory is detected and
> reported to operator? Or is it OK that broken memory is undetected and
> mangles
> packets without us being aware?

The impression I got was that the memory doesn't die until the device
is power cycled.
Is that incorrect?

Lee