What the data for this category looks like
Glass yield rate-related market data is sourced from daily market monitoring conducted by a domestic building materials circulation association. The association releases full data from the previous trading day daily at midnight. Data is stored in structured JSON format. Each record includes fields such as report date, glass category, benchmark quote, regional division, inventory turnover days, and market supply and demand index. The unit for benchmark quote is yuan per weight box, and the unit for inventory turnover days is calendar day. All fields correspond to the market performance of a single category in a single region, and no percentage-based statistical indicators are included.
Constraints for Deployment and Upgrade
Separate application is required for third-party interface access permissions for glass market data. During deployment, configure a dedicated API key and interface whitelist. Without these, valid data cannot be pulled. The daily midnight scheduled update task must align with the data update rhythm. Set the task interval to 24 hours to avoid frequent calls triggering interface rate limits. Since the data includes multi-regional and multi-category fields, configure dimension filtering rules during deployment. This ensures only target regional and category market data is loaded. The quote unit for all glass categories is uniformly yuan per weight box. During upgrades, verify data parsing logic. This prevents conflicts with unit conversion rules for other building material categories. Additionally, retain compatibility with the format of historical daily report data. The upgrade process must not disrupt the reading path of existing data.
Configuration Settings
| Configuration Item | Recommended Value | Rationale |
|---|---|---|
DATA_SYNC_CRON | 0 2 * * * | Matches the daily midnight update rhythm of glass daily reports, avoids calling third-party interfaces during business peak hours |
API_KEY_WHITELIST | ["Building Material Monitoring Platform Exclusive Domain"] | Restricts data pulling to authorized third-party data interfaces only, prevents unauthorized calls |
PARSE_FIELD_MAPPING | {"base_quote": "price", "area": "area", "category": "category"} | Maps Chinese fields from third-party data to standardized fields recognizable by the system, adapts to yield calculation logic |
MAX_SYNC_RETRY_TIMES | 3 | Addresses temporary fluctuations in third-party interfaces, avoids data loss caused by a single failed synchronization |
DATA_SYNC_TIMEOUT | 600 seconds | Adapts to the large single-batch pull volume of glass data, reserves sufficient timeout time to prevent premature interruption |
RECALL_FILTER_RULE | {"category": ["float_glass"], "area": ["east_china_region"]} | Focuses on market data of target segmented markets, avoids irrelevant data interfering with yield calculation |
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 individual analysis, and it is recommended to test on your own samples before finalizing settings.
Three Common Mistakes
- Symptom: After a new deployment, calling the data interface returns 403 Forbidden, and market data cannot be pulled. Cause:
API_KEY_WHITELISTis not configured, and authorized third-party data interfaces are not added to the whitelist. - Symptom: The generated yield broadcast content lacks core fields and displays empty. Cause:
PARSE_FIELD_MAPPINGis configured incorrectly, and core fields such as benchmark quote and category from third-party data are not properly mapped. - Symptom: The container exits immediately after executing
docker run --gpus all -itd -p 7231:7231to start it. Checking logs shows missing configuration. Cause: Version 4.9.0 added a mandatory configuration item forDATA_SYNC_CRON, and the data synchronization service cannot start without it set.
How to Verify Successful Configuration
- View system scheduled task logs to confirm that the task corresponding to
DATA_SYNC_CRONhas executed at the daily midnight rhythm. - Call the data pull interface and check if the returned fields correspond one-to-one with the configuration in
PARSE_FIELD_MAPPING. - Check the container's environment variables to confirm that
API_KEY_WHITELISThas been configured with authorized third-party interface domain names. - Manually trigger a data synchronization and check if glass market data can be normally pulled and parsed without error messages.
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.