Tailoring Multithreaded System Dump Diagnostics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In multithreaded transaction processing environments, diagnosing system dumps is challenging due to the vast number of threads and the lack of specific diagnostic data for active or suspended tasks, requiring service programmers to manually gather and format information from multiple system domains.
Innovation Solution
A computer-implemented method and system that tailors diagnostic information by scanning system dumps to identify running, failed, or suspended tasks, collecting and formatting relevant state information and control blocks associated with these tasks, and presenting them in a structured format for efficient problem determination.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If system dumps are created in multithreaded environments to capture complete system state, then diagnostic completeness is improved, but data volume and complexity increase making analysis difficult
Solution Approach 1:
The patent segments the monolithic system dump into task-specific diagnostic information. It identifies individual tasks (running, failed, or suspended) and extracts only the state information and control blocks relevant to each specific task, rather than presenting all system state data. This segmentation reduces the complexity of diagnostic data while maintaining completeness for the specific failure scenario.
Solution Approach 2:
The patent extracts only the necessary diagnostic elements from the complete system dump. It takes out and presents specifically the state information and control blocks associated with identified tasks, filtering out unrelated system state data. This extraction process maintains diagnostic completeness for the failure analysis while reducing overall data volume and complexity.
2Measurement precision
If complete system state information is presented for all threads, then diagnostic accuracy is improved, but time to analyze and resolve issues increases
Solution Approach 1:
The patent performs preliminary actions by automatically scanning the system dump to identify tasks that were running, failed, or suspended at the time of the dump. It pre-processes the data to determine which tasks are relevant to the diagnostic inquiry, then presents only that filtered information. This preliminary identification and filtering action maintains diagnostic accuracy while significantly reducing the time required for analysis compared to manual review of complete system state.
3Measurement precision
If manual gathering of diagnostic information is performed, then data accuracy is improved, but productivity and resolution speed decrease
Solution Approach 1:
The patent implements self-service by automatically scanning system dumps, identifying relevant tasks, collecting state information, and formatting diagnostic output without requiring manual intervention. The system serves itself by autonomously performing the diagnostic analysis that previously required service programmers to manually gather and format information from multiple system domains, thereby maintaining data accuracy while dramatically improving resolution speed and productivity.
Data Source
AI summary
Tailoring diagnostic information specific to current activity of multiple threads within a computer system. A request to dump system state is received. A system dump is created, including main memory and system state information. The system dump is stored to a database. In response to a request to format the system dump, the system dump is loaded from the database, whereby a virtual memory image of system state at system dump time is created. The virtual memory image and system state information is scanned to identify tasks that were running, tasks that have failed due to an error, and tasks that were suspended at system dump time. State information and control blocks associated with the identified tasks are collected from the system dump and collated based on task number. The database is updated with a formatted system dump, including the state information and control blocks associated with the identified tasks.