Re: [cobalt-users] Qube3 - How hot can she get?

[Malcolm McLeary <mmcleary@xxxxxxx>]

Hi Guys,

on 10/8/02 11:40 PM, Tim Dunn wrote:

> How do you get it to dump the temperature?

On the Qube 3 in the Active Monitor section ... look at CPU.

on 11/8/02 1:54 AM, Bob G7 wrote:

> This is from sun cobalt:
> http://www.sun.com/hardware/serverappliances/qube3/specs.html
> 
> Operating Environment: 32o to 95oF (0o to 35oC) 10% to 80% humidity
> (non-condensing)

I'd suggest that its possible that the local ambient conditions are getting
beyond 35 degrees.  (i.e. the Qube3's new home is inadequately ventilated).

> So I would say you need to look at the fan(s) and make sure they are working
> properly, check that the air flow vents are not blocked.

The Qube3 itself is fine ... where it lives is likely to be a problem.

on 11/8/02 2:14 AM, Jeff Lasman wrote:

> My PC hardware with bios temperature warning has the threshold set at
> 60; while I've never seen a Cobalt product run that high, it should be
> right at the high end of acceptable.

Its interesting that Cobalt provide a facility to monitor the temperature,
but don't appear to do anything about it.

>> Will a high cpu temperatures cause a Qube3 to crash?
> 
> Could, though more likely it would subtly affect system life.
> 
> Even our RaQs with broken fans (we only find out about them when we
> visit the colo) never run near that high.

My home office environment is probably a deal warmer than your average
environmentally controlled computer room.

on 11/8/02 2:28 AM, Gerald Waugh wrote:

> I would also suggest looking at 'top' this system maybe working its a** off.

I was watching top.  The conditions were a little artificial ... I wanted to
test it.  I had slipped in a blank disk so it was actively mirroring, I also
activated the memory test, hence the disks were working hard, the CPU was
constantly only a few % idle and all 512 MB of RAM was in use.

on 11/8/02 2:32 AM, Claudio Terra wrote:

> I've been monitoring the temperature of my RaQ4s for more than a year
> now with a custom MRTG script.

That would be better than going back and forth in Active Monitor screens.

on 11/8/02 3:12 AM, Gerald Waugh wrote:

> We have the same processor running in conventional redhat systems and they DO
> crash when they get hot.

I performed the same experiment two weeks ago, but with a different software
load ... all sorts or processes stopped responding before it finally
crashed.  I thought it could have been the software so this time the Qube3
was freshly restored with all the patches + SSH + Memory Tester.

> It could be your fan?

The fan performance appears to be consistent with another Qube3 I have here.

> We have found that in many cases the heat-sink compound between the processor
> and the heatsink has deteriated, cleaning both surfaces and applying a
> industrial grade compound fixes the issue.

I believe the problem is where the Qube3 lives, but I'll have a look at the
heatsink issue.  I might have to run some cable and relocate the Qube3 to
the garage as my normal workstation is an Apple G4 Cube and the Qube3 is
simply too noisy.

Cheers,  Malcolm