What the data for this category looks like
Daily documents for this category include four types: instance change orders, backup and recovery records, performance diagnosis reports, and parameter adjustment records. Data sources are cloud database consoles, slow logs, monitoring metrics, and ticket systems. Instance change orders include fields for instance ID, change time, operator, and change content. Backup and recovery records include fields for backup ID, recovery time, target instance, and status. Performance diagnosis reports include fields for slow SQL statements, execution duration, and lock wait time. Parameter adjustment records include fields for parameter name, pre- and post-adjustment values, and effective time. The data update frequency is daily or real-time upon instance change.
What constraints these characteristics impose on the document parsing and chunking workflow
Significant format differences exist across multi-source documents, including structured PDF tables exported from consoles, plain text lines from slow logs, and rich text content from ticket systems. General-purpose parsers struggle to adapt to all formats. Data update frequency is high, and the size of single documents accumulates with changes, increasing the risk of parsing timeouts. Some documents contain sensitive operation and maintenance data, so chunking must avoid splitting across sensitive boundaries to prevent data leaks. Fields include unit information such as execution duration and memory usage. Associated context must be retained during chunking, otherwise the data cannot be used for subsequent risk assessment and troubleshooting.
How to set the configurations
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
chunkSize | 800–1200 characters | Matches the average associated field length of a single operation and maintenance document, avoiding loss of context after splitting |
customPdfParse | Enabled | Adapts to PDF documents exported from the cloud console, retaining the typesetting structure of tables and text |
PARSE_FILE_TIMEOUT_SECONDS | 300 seconds | Addresses the parsing time required for large backup and recovery records, preventing task interruption before completion |
UPLOAD_FILE_MAX_SIZE | 1000 MB | Adapts to the maximum size of a single operation and maintenance document exported from the cloud database |
chunkSplitter | Split by paragraph and heading | Matches the hierarchical structure of operation and maintenance documents, retaining the chapter logic of change orders and reports |
chunkSettingMode | Manual configuration | Adapts to the chunking needs of different types of operation and maintenance documents, covering structured and unstructured scenarios |
Three common mistakes
- Phenomenon: After uploading a performance diagnosis report containing external image addresses, the parsing result does not include image content, and the interface prompts "file parsing exception". Cause: The
customPdfParseconfiguration is not enabled, and the default parsing logic does not support pulling and embedding external image links. - Phenomenon: After locally deploying MinerU and configuring the API port, checking the enhanced PDF parsing function, the parsing task remains in pending status. Cause: The MinerU service address and port are not configured in FastGPT's environment variables, or the network between containers is not connected.
- Phenomenon: Parsing large backup and recovery records causes the task to terminate early. Cause:
PARSE_FILE_TIMEOUT_SECONDSis set too short, failing to cover the parsing time required for large documents.
How to confirm the configuration is correct
- Check that the FastGPT version is v4.17.0, and confirm that the
customPdfParseswitch in the configuration page is enabled. - Upload a PDF-format instance change order exported from the cloud console, and check whether the parsing result retains the original typesetting of tables and text.
- After configuring the MinerU service address, initiate a PDF parsing task, and check whether the task log shows MinerU call records.
- Manually adjust
chunkSizeto the specified range, upload a performance diagnosis report, and check whether the fragment length of the chunking result meets expectations.
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.