bmc_health_config.json

This file outlines the format for the supplemental health metric config that can be supplied by the platform to override the default in-code config from health_metric_config.cpp.

The json config may have following metric types -

• CPU
  • This indicates the total CPU utilization.
• CPU_User
  • This indicates the user level CPU utilization.
• CPU_Kernel
  • This indicates the kernel level CPU utilization.
• Memory
  • This indicates the total memory for the system, which is a constant metric and doesn't change.
• Memory_Available
  • This indicates the memory which is not used, in addition to buffered & cached memory that can be made available by OS depending on need.
• Memory_Free
  • This indicates the amount of un-used memory.
• Memory_Shared
  • This indicates the amount of memory being shared between processes within the system.
• Memory_Buffered_And_Cached
  • This indicates the amount of memory being used for caching and temporary buffers.
• Storage_RW
  • This indicates the amount of available storage space
• Storage_<xxx>
  • This indicates the amount of availble space for type depicted by <xxx> for the location backed by path parameter.

The metric types may have the following attributes:

• Window_size
  • This indicates the number of samples being used for threshold value computations.
• Path
  • The path attribute is applicable to storage metrics and indicates the directory path for it.
• Threshold
  • The following threshold levels (with bounds) are supported.
    • HardShutdown_Lower
    • HardShutdown_Upper
    • SoftShutdown_Lower
    • SoftShutdown_Upper
    • PerformanceLoss_Lower
    • PerformanceLoss_Upper
    • Critical_Lower
    • Critical_Upper
    • Warning_Lower
    • Warning_Upper
  • Threshold may have following attributes
    • Value
      • This indicates the percentage value at which specific threshold gets asserted.
        • For lower bound, the threshold gets asserted if metric value falls below the specified threshold percentage value.
        • For upper bound, the threshold gets asserted if metric value goes beyond the specified threshold percentage value.
    • Log -A boolean value of true/false depicts if a critical system message shall be logged when threshold gets asserted.
    • Target
      • This indicates the systemd target which shall be run when the specific threshold gets asserted.

Example:

    "CPU": {
        "Window_size": 120,
        "Threshold": {
            "Critical_Upper": {
                "Value": 90.0,
                "Log": true,
                "Target": ""
            },
            "Warning_Upper": {
                "Value": 80.0,
                "Log": false,
                "Target": ""
            }
        }
    }