What the data for this category looks like
Data for this category primarily comes from CMDB systems, cloud resource inventories, automated discovery tools, and procurement ledgers. Daily produced documents include configuration item lists, service component relationship topologies, change records, and asset inventory sheets. A single document typically contains fields such as configuration item ID, asset type, dependency chain, change time, and others. Common units include unit, piece, and day. Data update cycles are daily or weekly. Some real-time change records require synchronous updates.
What constraints do these characteristics impose on the document parsing and chunking link?
Documents for configuration management and CMDB maintenance include strongly dependent relationship fields. Retain associated context during chunking. Do not split content across dependency relationships. Such splitting will impact subsequent impact analysis and fault location. Frequently updated daily or weekly documents must adapt to fast parsing processes. This avoids timeouts and meets real-time update requirements after changes. Documents from different sources have significant format differences. These include structured table-style asset lists and documents with embedded topology diagrams. Support multi-format parsing to meet multi-source data collection needs for configuration items. The field order of configuration items and dependency relationships is fixed. Retain field association after chunking. This ensures complete matching of asset chains during subsequent retrieval.
Configuration Settings
The following configurations apply to FastGPT v4.17.0.
| Configuration Item | Recommended Value | Rationale |
|---|---|---|
chunkSize | 800–1200 characters | Adapts to the field length of configuration item lists and topology documents. Retains complete context of dependency relationships |
chunkSplitter | Semantic chunking | Avoids splitting content across dependency relationships. Matches the structured association characteristics of CMDB documents |
customPdfParse | Enabled | Compatibility with PDF documents containing embedded topology diagrams. Resolves issues with unrecognizable content in some PDFs |
UPLOAD_FILE_MAX_SIZE | 1000 MB | Adapts to the upload requirements of batch asset inventory sheets. Prevents large file upload failures |
PARSE_FILE_TIMEOUT_SECONDS | 600 seconds | Adapts to the parsing duration of large weekly-updated topology documents. Prevents timeout errors |
Three Common Mistakes
- Symptom: A
413 Request Entity Too Largeerror is returned after uploading a CMDB document. Cause:UPLOAD_FILE_MAX_SIZEis not configured, or its value is smaller than the actual uploaded file size. - Symptom: Empty results are returned when parsing PDF topology documents. Cause: The
customPdfParseconfiguration is not enabled. The default parsing engine cannot recognize text embedded in topology diagrams. - Symptom: Field splitting chaos occurs after parsing configuration item lists. Cause: The
chunkSizevalue is too small. It splits configuration item entries across associated fields.
How to Verify Successful Configuration
- Upload a CMDB asset inventory sheet under 1000 MB. Check that the parsing task status is completed.
- Enable
customPdfParse, then upload a PDF document with an embedded topology diagram. Check that the parsing result contains topology text content. - View the
chunkSizeconfiguration item. Confirm that the value matches the average field length of the currently parsed documents. - Test uploading a large weekly-updated topology document. Confirm that the parsing duration does not exceed 600 seconds.
The values provided are common starting points and should be evaluated against local 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.