FRAGATA: Semantic Retrieval of HPC Support Tickets via Hybrid RAG over 20 Years of Request Tracker History
Summary: arXiv:2604.13721v1 Announce Type: cross
The technical support team of a supercomputing centre accumulates, over the course of decades, a large volume of resolved incidents that constitute critical operational knowledge. At the Galician Supercomputing Center (CESGA), this history has been managed for over twenty years with Request Tracker (RT), whose built-in search engine has significant limitations that hinder knowledge reuse by the support staff.
Introduction
This paper presents Fragata, a semantic ticket search system that combines modern information retrieval techniques with the full RT history. The motivation behind Fragata stems from the need to improve the efficiency and effectiveness of retrieving past support incidents, which are invaluable for both current operations and future troubleshooting.
Challenges in Traditional Systems
Traditional search capabilities of RT are often inadequate due to various factors:
- Linguistic diversity: The system must handle queries in multiple languages.
- Typographical errors: Support staff may input queries with typos, leading to missed results.
- Varied phrasing: Different wording for the same issue can hinder search results.
Overview of Fragata
Fragata addresses these challenges through a hybrid information retrieval approach. The system’s architecture leverages advanced algorithms, allowing it to:
- Identify relevant past incidents regardless of query language.
- Mitigate the impact of typographical errors on search results.
- Adapt to varied phrasing, ensuring a more inclusive search scope.
System Architecture
The architecture of Fragata is designed to seamlessly integrate with CESGA’s existing infrastructure. Key features include:
- Incremental Updates: The system supports real-time updates without service interruption, ensuring that the latest data is always available for search.
- Computational Offloading: The most resource-intensive stages of the search process are offloaded to the FinisTerrae III supercomputer, optimizing performance and response time.
Preliminary Results
Initial testing of Fragata has yielded promising results, demonstrating a significant qualitative improvement over RT’s native search capabilities. Users have reported enhanced satisfaction with the search results and an increased ability to retrieve relevant past incidents quickly.
Conclusion
Fragata represents a substantial advancement in the field of semantic search for high-performance computing support. By overcoming the limitations of traditional systems, it enhances the operational knowledge management at CESGA, thus contributing to more effective support and maintenance of supercomputing resources.
Future work will focus on further refining the system, incorporating user feedback, and potentially extending its application to other domains within the field of technical support.