Diagnostic method, apparatus, electronic device, and storage medium
By acquiring and aggregating the call stack and performance metrics of the target application, and using similarity strategies and threshold comparisons, the system automatically divides abnormal and normal call stack sets, and combines function frequency to attribute faults. This solves the problem of low fault diagnosis efficiency in existing technologies and achieves efficient and automatic performance problem localization.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-03-17
- Publication Date
- 2026-06-26
AI Technical Summary
Existing technologies struggle to automatically and accurately pinpoint application performance issues to specific code logic or business scenarios that cause performance degradation, leading to a heavy reliance on human experience and post-event analysis for troubleshooting, resulting in low efficiency.
By acquiring multiple target call stacks and their performance metrics during the runtime of the target application, the similarity between target call stacks is determined using a preset similarity determination strategy. Based on the similarity, the call stacks are aggregated to form call stack clusters. By comparing the performance metrics within the clusters with preset thresholds, abnormal and normal call stack sets are automatically divided, and fault attribution is performed based on the frequency of function occurrence.
It achieves an automated, data-driven diagnostic loop from problem symptoms to code root causes, significantly improving the efficiency and accuracy of problem diagnosis, automatically locating key code functions, and reducing reliance on manual investigation.
Smart Images

Figure CN122285473A_ABST