HPE Cray EX standard EIA rack node power management is supported by the server vendor BMC firmware. The BMC exposes the power control API for a node through the node’s Redfish Power schema.
Out-of-band power management data is polled by a collector and published on a Kafka bus for entry into the Power Management Database (PMDB). Access to the data stored in the PMDB is available through the System Monitoring Application (SMA) Grafana instance.
Power limiting of a node must be enabled and may require additional licenses to use. Refer to vendor documentation for instructions on how to enable power limiting and what licenses, if any, are needed.
CAPMC only handles power limiting of one hardware type at a time. Each vendor and server model has their own power limiting capabilities. Therefore a different power limit request will be needed for each vendor and model that needs to have its power limited.
cray-hms-smd) >= v1.30.16cray-hms-capmc) >= 1.31.0cray CLI >= 0.44.0See the CAPMC Deprecation Notice for more information.
get_node_energy (Deprecated)get_node_energy_stats (Deprecated)get_system_power (Deprecated)The Redfish API for rack-mounted nodes is the node’s Power resource which is presented by the BMC. OEM properties may be used to augment the Power schema to provide additional power management capabilities.
CAPMC power limiting controls for compute nodes can query component capabilities and manipulate the node power constraints. This functionality enables external software to establish an upper bound, or estimate a minimum bound, on the amount of power a system or a select subset of the system may consume.
CAPMC API calls provide means for third party software to implement advanced power management strategies using JSON data structures.
The rack-mounted compute nodes support these power limiting and monitoring API calls:
get_power_cap_capabilitiesget_power_capset_power_capIn general, rack-mounted compute nodes do not allow for power limiting of any installed accelerators separately from the node limit.
Power limit control will only be valid on a compute node when power limiting is enabled, the node is booted, and the node is in the Ready state as seen via the Hardware State Manager (HSM).
Get Node Power Control and Limit Settings
ncn-m001# cray capmc get_power_cap create –-nids NID_LIST --format json
Return the current power limit settings for a node and any accelerators that are installed. Valid settings are only returned if power limiting is enabled on the target nodes.
ncn-m001# cray capmc get_power_cap create --nids 4
{
"e": 0,
"err_msg": "",
"nids": [
{
"nid": 4,
"controls": [
{
"name": "Chassis Power Control",
"val": 500
}
]
}
]
}
Get Power Limiting Capabilities
ncn-m001# cray capmc get_power_cap_capabilities create –-nids NID_LIST --format json
Return the min and max power limit settings for the node list and any accelerators that are installed.
ncn-m001# cray capmc get_power_cap_capabilities create --nids 4 --format json
Expected output looks similar to the following:
{
"e": 0,
"err_msg": "",
"groups": [
{
"name": "3_AuthenticAMD_64c_244GiB_3200MHz_NoAccel",
"desc": "3_AuthenticAMD_64c_244GiB_3200MHz_NoAccel",
"host_limit_max": 0,
"host_limit_min": 0,
"static": 0,
"supply": 900,
"powerup": 0,
"nids": [
4
],
"controls": [
{
"name": "Chassis Power Control",
"desc": "Chassis Power Control",
"max": 900,
"min": 61
}
]
}
]
}
Set Node Power Limit
ncn-m001# cray capmc set_power_cap create --nids NID_LIST --control CONTROL_NAME VALUE --format json
Set the total power limit of the node by using the name of the node control. The power provided to the host CPU and memory is the total node power limit minus the power limits of each of the accelerators installed on the node.
ncn-m001# cray capmc set_power_cap create --nids 4 --control "Chassis Power Control" 600
Expected output looks similar to the following:
{
"e": 0,
"err_msg": "",
"nids": [
{
"nid": 4,
"e": 0,
"err_msg": ""
}
]
}
Multiple controls can be set at the same time on multiple nodes, but all target nodes must have the same set of controls available, otherwise the call will fail.
ncn-m001# cray capmc set_power_cap create \
--nids [1-4] --control "Chassis Power Control" 600
Expected output looks similar to the following:
{
"e": 0,
"err_msg": "",
"nids": [
{
"nid": 1,
"e": 0,
"err_msg": ""
},
{
"nid": 2,
"e": 0,
"err_msg": ""
},
{
"nid": 3,
"e": 0,
"err_msg": ""
},
{
"nid": 4,
"e": 0,
"err_msg": ""
}
]
}
Remove Node Power Limit (Set to Default)
ncn-m001# cray capmc set_power_cap create --nids NID_LIST --control CONTROL_NAME 0 --format json
Reset the power limit to the default maximum. Alternatively, using the max
value returned from get_power_cap_capabilities may also be used. Multiple
controls can be set at the same time on multiple nodes, but all target nodes
must have the same set of controls available, otherwise the call will fail.
ncn-m001# cray capmc set_power_cap create --nids 4 --control "Node Power Limit" 0
Expected output looks similar to the following:
{
"e": 0,
"err_msg": "",
"nids": [
{
"nid": 4,
"e": 0,
"err_msg": ""
}
]
}
Enable Power Limiting
ncn-m001# curl -k -u $login:$pass -H "Content-Type: application/json" \
-X POST https://${BMC}/redfish/v1/Chassis/Self/Power/Actions/LimitTrigger \
--data '{"PowerLimitTrigger": "Activate"}'
Deactivate Node Power Limit
ncn-m001# curl -k -u $login:$pass -H "Content-Type: application/json" \
-X POST https://${BMC}/redfish/v1/Chassis/Self/Power/Actions/LimitTrigger \
--data '{"PowerLimitTrigger": "Deactivate"}'