What the data for this category looks like
This category’s data primarily comes from stress testing platforms, APM monitoring systems, database slow logs, and production traffic baselines. Daily produced documents include stress testing plans, capacity evaluation reports, stress testing data records, and optimization suggestion sheets. Data update frequency is monthly or upon major version releases. Data fields include throughput rate, response latency, resource usage rate, peak request volume, and more. Corresponding units are typically requests/second, milliseconds, percentage, and times/minute.
What constraints do these characteristics impose on the citation sources and traceability link
Since data sources include multiple systems such as stress testing platforms and APM monitoring, the citation traceability process must clearly distinguish the recall scope of different data sources to avoid mixing irrelevant data. Data updates occur primarily monthly or at major version nodes, so the set of cited data sources must be fixed to prevent citation failures caused by temporary changes. The layered structure of documents requires matching content at the corresponding hierarchy during citation. Specific units of fields must be retained in citations to ensure readability and accuracy of performance indicators. Additionally, stress testing data has strong timeliness, so the range of historical data cited must be limited to avoid confusion of indicators across versions.
How to configure settings
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
| Maximum Citation Token Count | 800–1200 token | Matches the context processing limit of FastGPT v4.17.0, and covers the typical block length of stress testing reports and data records |
| Citation Data Toggle | Enabled | Ensures output content is bound to corresponding performance testing data to meet traceability requirements |
| Citation Details Toggle | Enabled | Retains field units and original values to avoid misinterpretation of performance indicators |
| Single Recall Limit | Top 3 entries | Matches the monthly updated stress testing data volume, avoids redundant information interfering with test case analysis |
| Data Source Filter | Only stress testing platforms, APM monitoring | Excludes non-performance testing data to ensure cited content accurately matches the business scenario of this category |
Three common misconfigurations
- Output only shows citation links without body content. This occurs because the citation data toggle is not enabled, and only data source identifiers are returned without loading actual performance testing data.
- Cannot select performance testing documents in the knowledge base as variables. This occurs because data source filter rules are not configured, and the system does not recognize the exclusive document library for this category.
- The AI fails to cite queried stress testing results. This occurs because the single recall limit is set too high, mixing non-current test case data from historical versions, leading to matching failures.
How to verify proper configuration
- Run a single performance test case generation task, check if the output includes specific performance indicators such as throughput rate and latency, along with their corresponding units.
- Review the citation source identifiers in the output content, confirm that only documents from specified data sources such as stress testing platforms and APM monitoring are included.
- Cross-check the update time of cited content to ensure it matches the current monthly or major version data set.
- Adjust the single recall limit parameter, verify that the number of cited entries in the output conforms to the set value range.
The values provided are common starting points and should be measured against the reader's own samples.
Question material comes from public community discussions. Configuration values are common starting points and should be measured against your own samples. Verified on 2026-10-19.