What this category of data looks like
This category of data mainly comes from VPC planning documents exported from cloud platform consoles, security group rule lists, load balancer configuration files, network topology diagrams, traffic monitoring reports, and architecture design documents submitted by customers. Daily generated documents include single-change configuration sheets, network segment planning tables, and health check policy descriptions. The data update frequency syncs with each configuration change or ticket delivery. The core fields of a single data entry include network segment range, port mapping rules, health check thresholds, and associated cloud resource IDs. There are no unified unit prefixes. Some fields require parsing based on the original format of the cloud vendor console.
What constraints these characteristics impose on the knowledge base retrieval and recall link
The data sources for this category are scattered across multiple formats including cloud platform consoles, network topology diagrams, and traffic monitoring reports. Some content consists of unstructured configuration fragments or screenshots. This requires the retrieval link to adapt to parsing logic for multi-source heterogeneous data. Data updates trigger synchronously with each configuration change or ticket delivery. The incremental update frequency of the knowledge base index must match the pace of business changes to avoid recalling outdated network segment rules or health check configurations. Core fields have no unified standard units. Additional context information of cloud resources must be associated during retrieval to complete accurate matching. The content focuses on network connectivity and traffic allocation rules, which differs significantly from resource provisioning data for other cloud service categories. Targeted adjustments to the matching priority of recall results are required.
How to Set the Configurations
| Configuration Item | Recommended Value | Rationale |
|---|---|---|
similarity | 0.75–0.85 | Matches the precision requirements for cloud network configurations, avoids recalling irrelevant security group rules or network segment plans |
limit | Top 8 entries | The number of configuration items involved in a single ticket usually does not exceed 7. Retaining redundancy covers multi-rule verification scenarios |
searchMode | semantic | Adapts to unstructured network topology and configuration documents, prioritizes matching semantically related network segment and port rules |
usingReRank | Enabled | Reorders recalled configuration items by relevance to the ticket question, optimizes the retrieval result order for health check configurations |
datasetSearchUsingExtensionQuery | Enabled | Automatically expands network terminology in tickets, associates "load balancer" with related configurations such as "port forwarding" and "session persistence" |
collectionIds | Bind the dedicated cloud network configuration knowledge base | Isolates knowledge base data from other categories, avoids recalling configuration content related to cloud hosts |
Three Common Misconfigurations
- Retrieval results include content from non-target categories. Phenomenon: Retrieval results contain configuration entries for cloud host resource provisioning. Cause: The dedicated
collectionIdsare not bound, leading to cross-knowledge-base retrieval of data from unrelated categories. - Unable to add image datasets in a private deployment environment. Phenomenon: No image dataset configuration entry appears in the knowledge base upload interface. Cause: The private deployment version of FastGPT v4.17.0 does not include the image dataset module in the basic knowledge base configuration process by default. Manual enabling of extended configuration is required.
- Unable to view historical call logs for individual documents. Phenomenon: The interface only displays global knowledge base call records, and cannot filter call situations for a single document. Cause: The current version of the log query module does not offer single-document dimension filtering, and only supports global dimension statistics.
How to Verify Successful Configuration
- Upload a test load balancer configuration document, submit a test ticket containing "port forwarding rules", and verify that the retrieval results only include cloud network-related configuration content.
- View the knowledge base index update log, and confirm that the index update frequency matches the time requirements of business changes.
- Submit a test query containing vague terminology, such as "check the health policy of the load balancer", and verify that the retrieval results are reordered by relevance.
- Enter the knowledge base configuration page, confirm that
collectionIdsis bound to the dedicated cloud network configuration knowledge base, and no knowledge bases from other categories are associated.
The values provided here are common starting points, and should be evaluated against the user'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.