Power management support was introduced for the CRS platform from IOS-XR version 4.3.0 onward.
The implementation includes the tracking of the available power to the chassis and the power consumed as well as chassis power zone monitoring for fixed (TDI) and modular (Arctic) power systems. It is limited to a static checking against known worst case numbers and logging warnings and alarms. It also implements a user interface (CLI) to display the worst case power consumption and the power availability.
Powermanagement is available for all LCs (MSC and PLIM) on CRS-1 and CRS-3. It covers the 4,8 and 16 slot chassis including RPs, Swich Fabric Cards, Alarm Modules, Fan Tays and Fan Controller.
On the fixed (TDI) power system the monitoring is based per power zone.
Scope and Limitations
Power management is based on a static worst case power table and slot power data. For the power usage calculations, always worst case (full load on all modules) power consumption data from the internal tables are assumed. It can’t be configured.
The software will only issue warnings. Modules are allowed to function even if they exceed the calculated chassis power availability.
When total power consumption is calculated, the power consumption of all mandatory cards like RPs, Switch Fabric Cards, Fan Trays, Fan Controllers and Alarm Modules, will always be added to the total chassis power usage. Independent if they are physically present. This prevents exceeding the power budget if redundant mandatory modules are added.
SPAs ,XFPs etc. are also icluded in the calculated power consumption, independent if they are inserted.
For the modular (Arctic) power system, the total power availability is the sum of all power modules in both shelves minus one. This one power module is reserved to cover a single module failure. The software will issue a separately worded warning and alarm when the power consumption crosses this threshold.
A major alarm is generated from shelf_mgr process if the power consuption is deemed to exceed the available power budget. The alpha display is set accordingly:
Modular (Arctic) Power-system: PWR CRITICAL
Fixed (TDI) Power-system: ZONEX PWR CRITCL (X= power zone)
A minor alarm is set when the redundancy threshold is crossed.
The alpha display is set to PWR LOW.
On systems with the fixed power system the following syslog message is issued when the worst case power budget is exceeded in a Zone:
%PLATFORM-SHELFMGRV2-2-INSUFFICIENT_ZONE_POWER : Power allotted in zone 2 has exceeded the available zone power budget. Please check the 'show power' command to resolve this situation.
On systems with the modular power system the following syslog message is issued when the worst case power budget is exceeded:
%PLATFORM-SHELFMGRV2-2-INSUFFICIENT_RACK_POWER : Power allotted to cards in this rack has exceeded the available rack power budget. Please check the 'show power' command to resolve this situation.
The following message is printed on a system with the modular power system when the worst case consumption exeeds the capacity to provide shelf redundancy:
%PLATFORM-SHELFMGRV2-4-POWER_MODULE_REDUNDANCY_LOST : Rack power is now being allotted from all power modules.Power module redundancy is no longer available, a single power module failure might result in card power loss.
RP/0/RP0/CPU0:CRS-X(admin)#show power ?
allotted displays power consumption information
capacity displays power capacity information
summary displays a summary of the power information
RP/0/RP0/CPU0:CRS-X(admin)#sh power allotted location 0/0/cpu0
Zone 3 and 6 exceeds the worst case( full load on all modules) power capacity. But all modules will continue to work if the real power draw is below the available power capacity. If the the real power consumtion exceeds the available budget on a Zone the whole Zone is shut down. If the modules cannot be shifted accordingly to avoid exceeding the power capacity a migration to the modular power system has to be considered.