keyboard failure that doesn't seem to be hardware

[Paul Allen Newell]

On 10/10/2012 10:28 PM, Paul Allen Newell wrote:
> On 10/10/2012 6:35 PM, Kevin Fenzi wrote:
>> When this happens, can you try holding down the shift key for say
>> 15-20seconds? Do you see a notice then about 'slow keys' being disabled
>> and it starts working again?
>>
>> kevin
>>
>
> Kevin:
>
> [...]
>
> This test will be the first I will try when I start again tomorrow.
>
> Paul
>

It behaved all day yesterday and I was into my 7th hour today when, 
finally, it happened again. Top shows no cpu hogging (near idle as far 
as I can see) and /var/log/messages doesn't report anything.

I ran your test and sat on the Shift Key for over a minute. No notices.

Figuring that was that, I ssh-ed into the machine to salvage what I was 
doing before killing the session. That took about 20 minutes and, 
somewhere around the 15th minute when I was mousing in one of the dead 
shells to scroll my history to make sure I had got everything, I 
accidentally brushed a key and, lo and behold, the keyboard was alive. 
Everything that had been typed while testing its dead-ness was not 
there, so all the input was not buffered waiting to be processed (as 
though the characters never made it to the computer).

So now I have to consider that "something" (be it hardware or software) 
is causing the keyboard to not exist to the computer for an unknown 
period of time. On one of the earlier tests, I waited about 10 minutes 
before killing and hadn't gotten the keyboard back by then.

I will try Ed's idea of a different keyboard, but I do want to ask if 
this new "experiment result" indicates additional things to consider in 
trying to fix it.

Thanks,
Paul