Proliant DL360 G3 overheating

[Stel]

We have this problem with a Proliant DL360 G3 server when the server shuts down unexpectedly due to overheating detected by sensor number 2.
Did anybody else experienced this type of issue?
There are two AC units next to the rack so the ambient temperature is normal.
I would like to know if this is an issue related to the temperature sensor that incorrectly reports the temperature forcing the server to shut down or it is something else? Any help would be highly appreciate it.

[Pedro Albuquerque]

Is the room with good cooling system? Can you tell me with insight manager? What is the temperature inside the server?
See also how is the threshold temperature set.
I hope this can help.

[Leon Rosier · Accepted solution]

Check out the mounting of the processor's coolingplate. it probably doesn't make proper contact with the processor. It is a common thing with this type of machine so check it out thoroughly

Leon

[Stel]

Thanks guys for your suggestions. I will look into all of them and let you know if it helped.

[Stel]

Thanks Leon for the suggestion. Apparently it was the CPU cooling plate causing the overheating issues. The CPU temperature for both CPUs on this server was 15 degrees higher that on any other servers(same model)located in the same rack.
Thank you all guys for your suggestions. My problem is fixed.

[Jaap Westerbeek_1]

We have the same issue on our DL360 G3, both with the dual 2.4 proc and the dual 2.8 proc (the single proc servers don't seem to have this problem)

We run Redhat 2.1 AS and when heavily stressed the server boots.
The HP manager tells us it's overheating in zone 3 at 53 degrees Celcius.

We've already had the motherboards replaced of 3 servers. Those servers didn't have the problem after that.

We run our servers in a properly cooled datacentre. servers are stacked 4u and then 1u "breathingspace"

HP is going to send some guys to measure our conditions. I'll keep you posted on the results...

[Leon Rosier]

When replacing the systemboard one also has to reinstall the coolplates so if one is mounted wrong it will be resolved as well.
In general to resolve this problem:
Install the latest Rombios and check if sw3 of SW2 on the systemboard is set to the off position and check the mounting of both coolplates.

Leon

[Jaap Westerbeek_1]

What is meant by the coolplate ? Is it the grey pasty layer on the bottom of the processor ? That is already attache to the proc OEM.

Setting the dipswitches would only change the fan speeds of the system, so that sounds like a workaround to me.

We're flashin the bios now, and will stresstest more after that.

[Leon Rosier]

The issue is, that in some situations the coolingplate didn't make proper contact with the processor core. When you would look at it, it looked like it wasn't placed horizontal.

sw3 to off lets the system management driver take control over the fanspeed.
So it is important that there is a management driver installed.

Leon

[Jaap Westerbeek_1]

The coolingplate, is that the Grill that's sitting on the processor ? That locks firmly onto it with to metal bars and a lever, and has coolingpaste in between. There's only one way it fits, and once fitted it very sturdy.



Regarding the fan speeds : We don't want to install the agents as we are using a heavily tweaked system that is dedicated for a single purpose. The Bios should be able to measure the temperatures and control the fans properly.

A bios upgrade did not fix our problem.