HarmonyOS APMS Fault Metrics: Multi-Dimensional App Health Monitoring

The article introduces HarmonyOS Application Performance Monitoring Service (APMS) fault metrics, a cloud-based observability feature that collects real-time fault events from client devices and provides multi-dimensional statistical analysis for developers.

Accessing Fault Metrics

Fault metrics are accessible in AppGallery Connect under Development & Services > Quality > APMS > Fault Metrics. Alternatively, developers can enter from the AGC homepage APMS entry and switch to the data overview.

Fault Metrics Page — Four Core Functions

The fault metrics page provides four main capabilities:

Metric Query : Filter by time range, app version, system version, device model, and other dimensions. Trend charts display key indicators such as crash rate (crashes per launch), crash count, and crash device count (unique devices reporting crashes in the time window).

Version Comparison : Compare the current version against any other version using the same filter conditions, enabling release-over-release regression detection.

Dimension Distribution : Breaks down the selected metric by app version, system version, and device model, helping pinpoint which versions or hardware configurations exhibit higher fault frequencies.

TOP Problem Aggregation Query : Aggregates associated exception logs into a ranked problem list; clicking an entry navigates to the fault analysis page for root-cause investigation.

Supported Fault Metric Categories

The service covers multiple fault types including crashes, freezes (ANR), application terminations, out-of-memory (OOM) events, and resource leaks. Each metric supports the query, comparison, distribution, and aggregation workflow, turning raw fault data into actionable quality improvement steps.

By leveraging these capabilities, developers shift from passive user-complaint-driven detection to proactive, data-driven daily inspection and early hazard discovery.