Inlet Temperature

[idata]

Hi,

We are having a Intel WB server hardware.

Chassis:

Prod code : SR1690WB

S/N number: AZGD9340548

BOARD:

AZWK93000042 PBA E48472-305

We are trying to use Node manager to reduce the sever power consumption, I am trying to limit the power through power limiting policy and could achieve it.

Now I am trying to set inlet temperature policy and reduce the power accordingly.

For this i have to first check for the "Get Node Manager Capabilities" and getting the completion code as 83h – unknown Policy Type

Also tried to get the Get Node Manager Statistics for the inlet temperature and it gives completion code as 88h – Invalid Mode

And also tried to list all the sdr in BMC and found that there are two temperature sensors

Baseboard Temp | 27 degrees C | ok

Front Panel Temp | 23 degrees C | ok

But I could not find which one is inlet temperature.

May I know how to get the values from Node Manager for inlet temperature and limit power according to it?

[idata]

Wouldn't the inlet temperature be ambient room temperature?

[Edward_Z_Intel]

Normally the Front Panel Temp is considered as inlet temperature.

[DSilv11]

I am trying to understand your usage case.

You want to throttle your server based on the room tempeature rathar than on power comsumption?

Is this due to lack of cooling capacity?

[Emilio_C_Intel]

Dear Pradeer Kumar:

SR1690WB does not implement any of the thermal options via Node Manager firmware.

Notice, that the Node Manager specification allows this feature as an option for the Intel Intelligent Node Manager firmware to implement.

The designers of this platform disabled or did not implement that option for this particular design, and the reason for the invalid responses.

The reason SR1690WB choose to do not implement this option is because the Baseboard Management Controller on the design provides this inlet thermal temperature information. The thermal sensor reported on SR1690WB platforms is managed by BMC for the software features the Intel management software provide with your SR1690WB.

The thermal sensor for BMC can still be access by the Intel provided software or by Intel Data Center Management SDK (DCM) or via IPMI tools.

See Information below for Intel temperature reading via DCM. I am providing information as an example to retrieve thermal information as an individual or group level usage model.

Similar approach can be taken with the Intel Server management provided with the system for implementation of the usage models described below.

Used Case: Single Node Thermal Control

To prevent a system from overheating, you can create a thermal policy to throttle down the system power to minimum when it reaches a thermal threshold you set.

Right click on the node and select policy, fill in the description and choose Minimum Power on Inlet Temperature trigger for Policy Type, make sure you have the correct node this policy is being applied to and enter the threshold value then check Policy Enabled to turn it on.

Note: For this test, you could set the threshold lower than the current temperature reading so it will trigger the policy and watch the system power decreases.

Figure 7. Thermal Policy Configuration.

Figure 8. Power level of URB001 has been brought down to minimum due to thermal threshold is reached.

Use Case: Group level Power and Thermal Monitoring

In DCM, you can create a group containing systems that share the same power circuit where total aggregate power consumption can be monitored.

Go to Configuration at the menu bar, select Add Group, assign a name to the group and Type could be Rack, Logical Group, DataCenter, Row, or Room. Once the group is created, add all the servers into the group and begin monitoring the total Power Usage of the servers in a group.

Figure 9. Create a Rack Group

Use Case: Group level Thermal Control

With thermal control, you can keep the system to becoming overheated by setting the thermal threshold.

Create a policy type of Minimum Power on inlet temperature trigger and set the temperature threshold level. Please note that the measurement is in Celsius.

Figure 19. Create thermal control policy at the Group level.

Once the policy is applied, you can see the total Demo group power consumption being brought down to idle until the current temperature is dropped below the thermal threshold of 20 degree Celsius.

Figure 20. Power is throttled down to idle to allow systems to cool down below thermal threshold.

At this point all the nodes being managed by DCM is working properly and we have successfully validated of their Node Manager Power and Thermal control functionalities

Sincerely,

Emilio Cañas

Senior Platform Engineer

Intel Architecture Systems Integration

Intel Corporation.