NVMe SMART log
The NVMe SMART log is the standardized log page containing controller health indicators and cumulative usage, endurance and error counters.
Log 02h is retrieved with Get Log Page. Critical Warning reports current conditions; Percentage Used estimates endurance consumed; error and power-loss counters count different events. Field names and warning bits depend on the implemented revision. [1]
Base 2.3 defines no namespace-specific information: supported per-namespace requests return the controller information. Do not sum repeated controller telemetry across namespace paths. [1]
Reading log 02h neither executes nor proves support for Device Self-test; its progress and results use log 06h. [3] Temperature-exposure counters use minutes, HCTM timers seconds; Warning Composite Temperature Time may include configured hysteresis. [1]
For JSON collection, smartctl --json=v requests KEY_s decimal strings for potentially large counters. KEY_le can be shorter than 16 bytes; the pinned zero-value serializer makes no array assignment. Missing companions are not a device reporting status. [4] [5]
A Number-only parse can lose unit increments; BigInt(value) cannot restore them. A source-aware reviver can use the original token. [6] [7]
Prometheus scalar samples use float64 even after exact decoding. Retain exact raw records when unit increments matter. Putting a changing count in a label creates a new series per distinct value. [8]
Relations
- related
- S.M.A.R.T., NVMe, NVMe Percentage Used, NVMe Available Spare, NVMe Critical Warning, NVMe Data Units Written, NVMe Unexpected Power Losses, NVMe Media and Data Integrity Errors, NVMe Error Information Log Entries, NVMe Controller Busy Time, NVMe Composite Temperature, NVMe Device Self-test, NVMe Device Self-test log, NVMe Critical Composite Temperature Time, NVMe Host Controlled Thermal Management, NVMe Warning Composite Temperature Time, NVMe base specification, NVMe Error Information Log, NVMe Persistent Event Log, NVMe Data Units Read, NVMe Host Read Commands, NVMe Asynchronous Event Request, NVMe Commands Supported and Effects
Limitations
- A successful log read supplies telemetry, not a guarantee of data integrity or future availability; a zero field must be interpreted according to that field's reporting rules. [1, 2]
Citations
- NVM Express Base Specification 2.3, section 5.2.12.1.3, Figure 210, printed pp. 213-217: SMART / Health Information
- libnvme: nvme_get_log_smart, SMART / Health Information retrieval
- NVM Express Base 2.3: sections 5.2.5, 5.2.12.1.7 and 8.1.8; Figures 174-177, 215-216 and 328
- smartmontools pinned manual: --json=v exact integer companions
- smartmontools pinned JSON implementation: set_unsafe_uint128 and compact little-endian arrays
- RFC 8259, section 6: JSON numbers and integer interoperability
- ECMAScript: InternalizeJSONProperty and original source text in reviver context
- Prometheus data model: float64 scalar samples and time-series label identity