What the data for this category looks like
Plastics and rubber market data is sourced primarily from listed futures products on the Shanghai Futures Exchange, plus daily quotes from domestic bulk commodity spot traders. Update frequency: spot quotes update once per hour on trading days, futures data pushes in real time during trading hours, and daily settlement reports generate after market close. Data is provided as structured tables with these fields: product name, trading market, daily opening price, daily closing price, settlement price, trading volume, position volume, spot quote. Units include yuan/ton, lots, tons, and others.
Constraints for conversation logging and auditing
High-frequency, multi-channel data updates require conversation logs to accurately record the timestamp of each query and the corresponding data version. This prevents tracing deviations during audits caused by data updates. Differences across multiple fields and units (such as futures trading volume measured in lots, spot trading volume measured in tons) require logs to fully retain field and unit information, to avoid audit confusion. Data differences across trading time periods require logs to distinguish pre-market, during-market, and after-market query records. This ensures auditors can restore the data source state at the time of the query. Multi-source data requires channel labels in logs, to meet compliance audit tracing requirements.
How to configure settings
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
LOG_RETENTION_DAYS | 90 days | Bulk commodity industry audits typically require at least 90 days of business operation and data query records |
RECORD_DATA_SOURCE | Enabled | Plastics and rubber market data comes from multiple futures and spot channels. Audits require clear labeling of the data source for each log entry |
LOG_FIELD_INCLUDE_LIST | variety name, trading market, daily closing price, data source, query time | Covers core market fields and tracing information required for audits, and avoids redundant logs that consume storage |
LOG_EXPORT_MAX_SIZE | 500 MB | Supports bulk export of multi-day market log scenarios, and prevents excessively large export files that cannot be transferred |
SESSION_STORAGE_DIR | /data/fastgpt/sessions | For Docker deployments, mount session storage to a persistent directory to prevent loss of historical conversation records after container restarts |
SESSION_PERMISSION_MODE | Bind user identifier | Commercial quotes for plastics and rubber involve industry privacy. Conversation records must be restricted to authorized users only |
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 case-by-case analysis. It is recommended to test on internal samples before finalizing configuration settings.
Three common configuration errors
- Symptom: After sharing a market quote bot via web link, external visitors see a "No permission to access this conversation record" prompt when attempting to view historical conversations. Cause:
SESSION_PERMISSION_MODEis not configured to a permission mode compatible with the sharing scenario, or the shared link does not carry a valid session access credential. - Symptom: After restarting a Docker-deployed FastGPT instance, some historical conversation logs and temporary cache files are lost, and audit tracing cannot be completed. Cause: The session storage directory is configured as a temporary internal container path, and is not mounted to the host’s persistent storage directory.
- Symptom: Exported conversation logs do not label the source channel of plastics and rubber market data. Auditors cannot verify data authenticity. Cause: The
RECORD_DATA_SOURCEconfiguration item is not enabled, and the data source field is not included in the log recording scope.
How to confirm configuration is applied correctly
- Access the FastGPT backend log management page, check for market query logs that include data source and trading market fields. Verify that log fields match the configured
LOG_FIELD_INCLUDE_LIST. - Create a conversation and query plastics and rubber market data. Generate a shared link, then use accounts with different permissions to access the link. Confirm that the
SESSION_PERMISSION_MODEconfiguration takes effect. - Review the Docker container’s mount configuration, confirm that the session storage directory is correctly mounted to the host’s persistent path, to prevent log loss after container restarts.
- Trigger a bulk log export operation, check that the export process has no abnormal errors, and that the exported file format and fields meet expected standards.
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.