What the data for this category looks like
Object storage and backup service business data primarily comes from object storage consoles, backup systems, access logs, and customer policy tables. Daily maintenance documents include bucket configurations, backup policies, recovery records, and lifecycle rules. Data updates occur either via daily synchronization or real-time updates when policies change. Fields cover bucket capacity, backup task status, recovery duration, permission policy entries, and more. Some fields are structured numerical values. Others are unstructured text.
What constraints do these characteristics impose on reranking and hybrid retrieval workflows
Data for this category is scattered across multiple systems. It includes two types of documents: configuration-related and record-related. Update cycles differ between the two types. Hybrid retrieval must recall both unstructured text such as bucket configurations and backup policies, and structured data such as recovery records and access logs. This creates clear requirements for retrieval patterns.
Different data types have significantly different business weights. For example, the relevance priority of permission policies and recovery records is higher than that of regular bucket logs. The reranking process therefore requires targeted adjustments to sorting weights. Additionally, indexes must be synchronized promptly after data updates. This avoids retrieving expired policy or task information.
How to configure settings
The following configurations apply to FastGPT v4.17.0 and later versions.
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
usingReRank | Enabled | Adapts to the relevance sorting needs of multi-source data for object storage and backup services. It filters low-relevance bucket logs or redundant configurations. |
similarity | 0.75–0.85 | Balances retrieval recall rate and precision. It suits business data with large text variance such as permission policies and recovery records. |
limit | Top 10 entries | Controls the number of returned results. It avoids displaying excessive redundant bucket access logs during customer technical support. |
searchMode | hybrid | Enables hybrid retrieval mode. It recalls both structured backup records and unstructured configuration documents. |
Three common configuration errors
- Enabling reranking in the application results in no search results, but standalone vector retrieval works normally. Troubleshooting steps: Verify that the
AIPROXY_API_ENDPOINTandAIPROXY_API_TOKENenvironment variables for the App/Admin server are unconfigured or misconfigured. These settings are required to call the AI Proxy service. Incorrect configuration may cause the service to fail to call normally and return empty results. - Reranking results differ minimally from pure vector retrieval results. Troubleshooting steps: Verify that the
similaritythreshold is not set too high. An overly high threshold may only recall a small number of highly relevant data. This prevents the reranking process from effectively adjusting sorting. - Search results include expired lifecycle rules. Troubleshooting steps: Verify that a reasonable index synchronization trigger timing is configured. Missing or improper timing may fail to update the latest policy data in the index in a timely manner.
How to confirm successful configuration
- Check the environment variables on the App/Admin server. Confirm that
AIPROXY_API_ENDPOINT(AI Proxy service root address, do not append paths such as /v1, not a single-model address) andAIPROXY_API_TOKEN(AI Proxy administrator token, consistent with server configuration and non-empty, not an application API key or similar) are correctly configured and non-empty. - Initiate an in-app reranking retrieval request. Enter a typical customer question such as "How to recover an expired backup task". Verify that returned results include backup policies and recovery records with reasonable sorting.
- Modify the lifecycle rule of one bucket. Wait 1 hour, then initiate a retrieval. Confirm that returned results include the updated rule content.
- View the AI Proxy service logs. Confirm there are no timeout or authentication failure errors for reranking model calls.
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.