[rescue] Octane woes

[David Holland]

Yeah.. that's what mine did (does?)

If you connect to the serial port, you'll get a bit of "junk" out of it 
when you power the system up.   "DejaNews"ing around will eventually 
turn up a article that has the bits in it, as the "junk" is fairly 
consistant.

That article is what I ended up doing, and now am kinda keeping an eye 
out for a new IP30.

Anyone know specifically which rev's had the problem, and perhaps more 
importantly, wanna sell a newer one? :-)

David


Jonathan C. Patschke wrote:

> On Thu, 6 Mar 2003, Curtis H. Wilbar Jr. wrote:
> 
> 
>>One thing I did do... is on the XIO (I think that is what it is) where
>>the graphics card is... I touched the funky connector... the one your
>>not supposed to touch.... doubt that could cause this...
>>
> 
> No, really, that -can- kill the board, just like the documentation says.
> You should still get some activity, though.  I forget what the guys at
> SGI said to clean those connectors with.  I think a horsehair brush was
> their recommendation.  Forced air (even "computer grade" canned air) is
> right out.
> 
> 
>>XIO boards and powering up... the light never comes on the front... so
>>it acts more like the cpu board is either not getting power, or has gone
>>dead....
>>
> 
> Is the CPU-board fully-seated?  What revision IP30 do you have?  At
> least one revision has a fatal flaw in the "Heart" chip (the chip under
> the big heat sink in the center of the board) that causes the system to
> appear dead.  If my suggestion in the last paragraph doesn't work try
> this:
> 
>   1) Pull the IP30.  Place it on a flat, sturdy surface that supports
>      the -board-, not the cage it's in (like a book or something of
>      similar size).
>   2) Use your thumbs to apply even pressure to the heatsink in the middle
>      of the board.  You want about 20 to 30 pounds of pressure, which is
>      why you want to support the -board- and not the frame.
>   3) Try reinserting the IP30.  If the system boots, you have one of
>      those defective boards.
> 
> I posted about this particular problem in comp.sys.sgi.hardware about 18
> months ago.  Here's a link:
> 
> http://groups.google.com/groups?threadm=3BF7E01B.184F05E4%40celestrion.theobvious.net
> 
> And when I got it (briefly) working:
> 
> http://groups.google.com/groups?threadm=3BF92AC9.232F148B%40celestrion.theobvious.net
> 
> I ended up replacing that board.
> 
> 
>>Are those funky connectors touchy....  they don't give me the warn and fuzzy
>>fealings of something reliable and trustworthy.... but I have no experience
>>and am going by physical impressions... not electrical characteristics....
>>
> 
> They're very reliable, so long as they stay -CLEAN-.  Sun uses them for
> attaching processors in their Exx00 series systems.  The most obvious
> benefit is the sheer number of "pins" you can have in such a small
> space.
> 
> 
>>I hope this thing isn't dead now... <sigh>...
>>
> 
> Try yanking the video card.  Pull it out just so it's disconnected from
> the frontplane, but so that there's still enough of the cardcage inside
> to maintain air pressure.  Power on the system.  You should get a
> blinking red lightbar.  If you do, you know what's dead.