What this category of data looks like
Coke due diligence data is sourced from three main channels: Dalian Commodity Exchange public delivery standards, monthly supply and demand reports from the China Coking Industry Association, and real-time coastal port inventory ledgers.
Data update schedules fall into three categories:
- Delivery grades and physical and chemical indicators are synchronized with exchange market data daily.
- Weekly production and port inventory data is updated every 3 days.
- Monthly supply and demand reports are released before the 10th day of each month.
Each individual due diligence document has a fixed structure: core indicators page, regional supply and demand page, and price trend page. Fields include ash content, volatile matter, sulfur content, crushing strength M40, abrasion resistance M10, transaction price (yuan/ton), port inventory (10,000 tons), and more. All physical and chemical indicators are labeled with units corresponding to national standard detection methods.
Constraints on Deployment and Upgrade
The multi-source and varied update schedule characteristics of coke due diligence data impose multiple constraints on deployment and upgrade processes. First, differences in update frequencies across data sources require deploying differentiated scheduled synchronization tasks. A fixed-interval trigger logic cannot be used uniformly. Second, fields include national standard physical and chemical indicators and exclusive units. Strict field validation and mapping rules must be preset in the data parsing module to prevent non-standard data from entering the knowledge base. Third, differences in document structures across cycles require adapting multiple parsing templates during upgrades. A single parsing rule cannot be reused. Fourth, fixed standards for coke delivery grades require presetting field matching logic during deployment to ensure external imported data aligns with national standard indicators. Fifth, the large content volume of individual monthly due diligence documents requires configuring vector storage parameters adapted for long texts to avoid truncation of core information.
Configuration Settings
| Configuration Item | Recommended Value | Rationale |
|---|---|---|
MONGODB_VERSION | MongoDB 6.0 or 7.0 | Compatible with the version range officially supported by FastGPT, can stably store vector indexes and metadata for multi-source coke data |
PARSE_FILE_TIMEOUT_SECONDS | 600 seconds | Covers the full parsing duration of individual monthly coke due diligence documents, prevents mid-parsing timeout interruptions for long documents |
UPLOAD_FILE_MAX_SIZE | 2000 MB | Adapts to the maximum compressed volume of individual monthly coke due diligence reports, meets full document upload requirements |
PARSE_SEGMENT_LENGTH | 1000–1200 characters | Preserves contextual connections between coke physical and chemical indicators and supply and demand data, prevents separation of core indicators and corresponding data |
VECTOR_STORE_BATCH_SIZE | 50 entries | Balances storage efficiency and index stability for multi-field coke data, prevents single-batch storage timeouts |
API_CONFIG_SYNC_MODE | Triggered via configuration file | Adapts to regular update requirements for third-party API keys and addresses for coke data, prevents frequent interface call errors |
The parameter values provided on this page are common starting points for configuration. Actual values are affected by material form, data volume, and business rules. Specific issues require targeted analysis. It is recommended to test on your own samples before finalizing settings.
Three Common Misconfigurations
- Issue: After running a cross-version upgrade script, the system returns a "400 Incompatible Version" error. Cause: Vector storage metadata for MongoDB was not exported in advance. Direct execution of the upgrade script results in incompatible index structures between old and new versions.
- Issue: After importing coke due diligence documents, core fields such as ash content and crushing strength M40 display as empty. Cause: Exclusive mapping rules for
PARSE_FIELD_MAPPINGwere not configured. The parsing module cannot recognize field naming formats unique to the coke industry. - Issue: After adjusting third-party API configurations, data synchronization tasks continue to fail. Cause: The binding relationship between API authorization keys and interface addresses was not updated synchronously. This results in a mismatch between configuration items and actual call parameters.
How to Confirm Proper Configuration
- Log in to the FastGPT system configuration page, and verify that the
MONGODB_VERSIONconfiguration item matches the actual MongoDB version of the deployment environment. - Upload a standard monthly coke due diligence document, check that the parsing task status is marked as completed, and confirm that the parsing duration falls within the preset timeout threshold range.
- Trigger a third-party API synchronization task, check the synchronization logs for core coke indicator fields, and confirm that the field mapping rules are active.
- Run the pre-check command for the upgrade script, check that the logs contain no version conflict or configuration missing prompts, and confirm the compatibility of the upgrade process.
Question material comes from public community discussions. Configuration values are common starting points and should be measured against your own samples. Verified on 2026-09-14.