Disk Health Metrics

Starting with Longhorn v1.11.0, disk health monitoring metrics are available for both the V1 and V2 data engines. These metrics provide insights into disk health status.

Note:

• Health data is collected periodically (every 10 minutes) by Longhorn’s disk monitor.
• Some virtualized or cloud environments (e.g., AWS EBS) may not expose full SMART data, resulting in zero values for certain attributes.
• Available health attributes may vary depending on the underlying disk type and hardware.
• The full set of collected health data is available in the nodes.longhorn.io Custom Resources (CRs).

Data Sources

• V1 Data Engine:
  • Health data is collected using the SMART monitoring tool (smartctl).
• V2 Data Engine:
  • NVMe disks: Health data is retrieved through SPDK.
  • AIO disks: Health data is collected using the SMART monitoring tool (smartctl).

Note:

Health data is sourced differently depending on disk type:

• V1 disks and V2 AIO disks: via SMART
• V2 NVMe disks: via SPDK

Available attributes and formats may vary by disk type and hardware. For details:

• SMART attributes: smartmontools documentation
• SPDK NVMe health data: bdev_nvme_get_controller_health_info JSON-RPC.

Health Attributes

The longhorn_disk_health_attribute_raw metric exposes raw attribute values with the following labels:

• attribute: Name of the attribute
• attribute_id: Attribute ID, when provided by the collection method.
• disk: Longhorn disk identifier.
• node: Name of the node.

Note:

SMART data may not be available on all platforms (particularly cloud providers). If SMART is not supported, health metrics will appear as 0.

References

• Related Github Issue #12016.