kempnerpulse.compute.health

kempnerpulse.compute.health(record, *, pcie_replay_rate_per_second, model_name)[source]

Return (health_label, rich_style) for one record.

Cascade, first match wins:

  • CRIT — a row-remap failure flag, or any uncorrectable remapped rows.

  • WARN — a PCIe replay rate that is present and positive.

  • HOT — a die or memory temperature at/above the model warning threshold.

  • OK — none of the above.

Parameters:
Return type:

Tuple[str, str]