What Data for This Category Looks Like
This category’s data comes from fault timelines, logs, monitoring metrics, and code commit records. Updates occur once per major incident or monthly. Daily documents include RCA reports, incident tickets, post-mortem minutes, and rectification tracking sheets. Document structures typically include fields such as incident occurrence time, root cause analysis conclusion, verification steps, rectification items, and tracking status. Fields adjust flexibly based on specific incident scenarios, with no mandatory unified units.
Constraints Imposed by These Characteristics on Knowledge Base Retrieval and Recall
This category has scattered, diverse data sources. Sources include structured monitoring metrics, semi-structured logs and code commit records, and unstructured post-mortem documents. The retrieval link must support cross-format content matching. Since updates occur once per major incident or monthly, high-frequency incremental synchronization does not need to be configured. Regular full updates can be used instead. Documents also contain dedicated fields such as root cause and rectification items. Retrieval must prioritize matching content strongly related to incident analysis to avoid irrelevant information interfering with root cause localization.
Configuration Settings
| Configuration Item | Recommended Value | Rationale |
|---|---|---|
similarity | 0.75–0.85 | This category requires prioritizing matching content strongly related to root cause analysis. This range filters low-correlation historical incident documents |
limit | Top 10 entries | Adapts to SRE teams' need to quickly review core cases, avoiding excessive results that increase information filtering costs |
usingReRank | Enabled | Multi-source data has diverse formats. Reranking improves the sorting priority of root cause-related content and optimizes retrieval result quality |
searchMode | Semantic search | Queries for this category are often vague incident phenomena. Semantic search matches implicit root cause associations |
datasetSearchUsingExtensionQuery | Enabled | Automatically expands retrieval keywords based on the user's incident description, covering more relevant historical documents |
collectionIds | IDs of fault analysis knowledge base collections associated with the current project | Limits retrieval scope to avoid searching irrelevant business knowledge base content |
Three Common Misconfigurations
- Symptom: An embedding error is returned in the interface when uploading RCA reports or fault logs. Cause: The format of multi-source data for this category is not adapted. Long text logs are not split into reasonable lengths, exceeding the processing limit of the embedding model.
- Symptom: An initialization error occurs when creating a fault analysis knowledge base using the FastGPT v4.17.0 deployment via docker-compose. Cause: The knowledge base storage directory is not correctly mounted in the deployment configuration, resulting in failed data writes.
- Symptom: When configuring the knowledge base search plugin, the specified fault analysis knowledge base cannot be selected via variables. Cause: The
collectionIdsparameter is not correctly configured, and the ID range of the target knowledge base is not limited.
How to Confirm Proper Configuration
- Upload a test post-mortem document for a fault, confirm no embedding error prompts are returned.
- Enter a query simulating a fault phenomenon, verify that retrieval results prioritize displaying historical documents related to root cause analysis.
- Check that configuration parameter values adapt to the multi-source data format and update rhythm of this category.
- Verify that the knowledge base search plugin configuration can accurately associate with the target fault analysis knowledge base collection.
The values provided are common starting points and should be evaluated against respective deployment 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.