New approaches in user centric job monitoring on the LHC computing grid : application of remote debugging and real time data selection techniques / by Tim dos Santos. 2011
Inhalt
- I Introduction
- Context: On High Energy Physics (HEP)
- Current research in HEP
- CERN and the LHC
- The Large Hadron Collider
- The ATLAS Experiment
- Data flow in ATLAS
- Real-time data reduction: Triggers
- Software in HEP
- Grid Computing
- Overview
- Definition of the term ``Grid Computing''
- Virtual Organisations
- Components and services of a Grid
- Security in the Grid
- The WLCG
- gLite Grid jobs
- WLCG software
- Conclusion
- II Job monitoring
- Overview
- The Job Execution Monitor
- Concept
- Architecture
- Acquisition of monitoring data
- User interface
- Deployment strategy
- Shortcomings of this version of the software
- Conclusion
- III Tracing the execution of binaries
- Concept and requirements
- Event notification
- Symbol resolving and identifier lookup
- Application memory inspection
- Publishing of the gathered data
- User code prerequisites
- Architecture and implementation
- Event notification
- Symbol and value resolving
- A victim-thread for safe memory inspection
- Resulting monitoring data
- Usage
- Stand-alone execution for custom binaries
- Integration into JEM
- Configuration and invocation
- Insertion of CTracer-data into JEMs data stream
- Augmentation of the JEM-Ganga-Integration
- Application for HEP Grid jobs
- Preparation of the user application
- Activation and configuration in Ganga
- Results and interpretation in an example run
- Performance impact
- Conclusion
- IV A real time trigger mechanism
- Concept and requirements
- Extendible chunk format for monitoring data
- Chunk backlog and tagging
- Inter-process communication in JEM revised
- Architecture and implementation
- General JEM architecture changes
- High-throughput shared ring buffer
- Triggers and event handling
- Memory management
- Application in JEM
- Testing
- Conclusion
- V Summary
- VI Appendices