[SUMMARY] sunfire 280R temperature difference between two CPUs

From: Ben K. <ben9coder_at_gmail.com>
Date: Mon May 08 2006 - 12:12:40 EDT
Thanks for all who replied - Eric Voisard, Ken Rossman, and Mark
Wiederspahn.  I quoted the responses below.

My original question was:

We have two SunFire 280R servers, but in one, the CPU temperature is like
> 142 F and 125 F, and in another, CPU temperature is 131F and 131F.
> I wonder why in the first one, the CPU temperature is not the same. I
> wonder
> if this is normal or not. I wonder if the cpustat or cputrack can be used
> to
> find a clue to this but am not sure which event to use.

It seems the difference in temperature is not abnormal and something I can
live with. I'd be more comfortable if I knew the reason, so I'll keep on
checking. I had no responses about how to check for difference in cpu
utilization, cpustat or cputrack.

Initially I precluded anything mechanical as cause, since there's another
server right below this in the same cabinet, which shows the same
temperature for both CPUs. And I certainly wouldn't have modified the
internal air path when I  opened the server's chassis last time. RSC didn't
show failures either. I'll look at the  internal when I get to reboot the
server next time and check for other component failures.

Ken Rossman <rossman@columbia.edu>

There's a pair of fans in the rear, if I am not mistaken.  If I were
to hazard a guess, one of these fans has failed.  Try launching the
Solaris Management Console and have a look at some of the pictures
of the system that should be in there.  They should show you where
some components may have failed.

Eric Voisard

All our 280R here do have such a difference in temperature between both
CPUs. If you open the case, you'll see that the fans' duct which blows
fresh air to the CPUs is placed assymetrically and CPU0 is slightly
better cooled than CPU1. Actually on a 280R here, CPU0 is 600C (1400F)
and CPU1 is 640C (147.20F).

Mark Wiederspahn <markw@ig.utexas.edu>

cpu temperature

This may help; sun-blade2000's are (almost) the same as 280r's but easier to
ask them what the temperature is. Depending on your cpu clock rate, you
look into the heatsink to cpu bonding; I saw that for some cpu boards there
was an
ultra-flat engineering revision; can't remember if it was sparc-III+ or
Sunsolve had the details. I assume you've looked at obvious air blockage
issues, etc.

I think the two cpu temps should be closer than you see. At least in my
cpu0 is always a bit hotter than cpu1, but not by that much.

Here are my temperatures; note that these systems have -much- different
ambient temperatures because they are in offices across three buildings.

900Mhz 280r #1:
System Temperatures (Celsius):
     CPU0    56
     CPU1    52
      RSC    30

900 Mhz 2804 #2:
System Temperatures (Celsius):
     CPU0    56
     CPU1    54
      RSC    30
sunmanagers mailing list
Received on Mon May 8 12:13:22 2006

This archive was generated by hypermail 2.1.8 : Thu Mar 03 2016 - 06:43:57 EST