What the data for this category looks like
Data for this category comes from registered process documents, raw material batch records, DCS systems, weighing equipment, LIMS, chromatography data systems, and QMS. Daily documents include batch production records, material balance ledgers, reaction and purification step records, cleaning validation records, impurity profile reports, deviation and change documents. Data is updated per production batch and per process step. Fields include batch numbers, reaction parameters, material feeding amounts, impurity test values, and yield data. Units cover mass, volume, concentration, and batch number-related content.
What constraints these characteristics impose during deployment and upgrade
Multiple heterogeneous data sources require configuring parsing adapters that support multiple document formats during deployment. Existing LIMS and chromatography system interface specifications must be connected in advance. High-frequency updates per batch and step require configuring storage parameters that support high-concurrency read and write operations. This avoids single-batch data write timeouts. The fine-grained data structure with multiple fields requires configuring sufficient field mapping rules during deployment. This ensures that ledger fields match the system storage structure. Compliance documents related to registration must be archived separately. Upgrades must retain access permissions for historical data. This prevents impacts on compliance traceability. The storage priority of core data such as impurity profiles and yields must be adjusted during deployment by reallocating storage node resources.
How to set the configurations
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
DB_MAX_LINK | 100–150 concurrent connections | Adapts to the concurrent demand of high-frequency read/write operations for multiple batches, and matches the volume of multi-source data access in API manufacturing |
MONGODB_URI | Replica set mode connection string | Stores unstructured documents such as impurity profiles and batch records. Replica sets ensure data reliability and high-concurrency read/write performance |
REDIS_URL | Cluster connection address with password authentication | Caches frequently queried batch yield and impurity test data. This reduces database read/write pressure |
STORAGE_S3_ENDPOINT | Enterprise-owned storage node address with bandwidth of no less than 100 MB/s | Archives large-volume chromatography data files and impurity profile reports. This prevents local storage overflow |
FILE_TOKEN_KEY | 32-byte random secret key | Ensures signature security for file upload and download. This prevents unauthorized access to document resources |
ROOT_KEY | Secret key generated according to enterprise permission rules | Controls system administrator permissions. This ensures access auditing for compliance documents |
Three common mistakes
- Issue: Server read/write load crashes after creating a knowledge base vector via the interface. This occurs daily on a scheduled trigger. Cause: The
DB_MAX_LINKparameter was not adjusted for high-frequency batch data updates. Excessive connection usage from single-batch writes leads to resource exhaustion. - Symptom: Plugin loading fails with a MongoServerError. This error alone does not establish a conflict between MongoDB 4.4.29 and FastGPT v4.17.0 plugin compatibility; inspect the complete logs, the database request issued by the plugin, and account permissions. The v4.17.2 deployment config documents 4.4.29 as a fallback for CPUs without AVX (see the official deployment configuration).
- Issue: Imported knowledge base backups do not distinguish between multiple files. Cause:
STORAGE_S3_ENDPOINTwas not configured to use an independent storage path. This causes backup files to be overwritten or merged.
How to confirm the configuration is complete
- Initiate a batch concurrent write test. Observe server CPU and memory usage. Adjust the
DB_MAX_LINKparameter until resource usage stabilizes. - Test the plugin loading process. Confirm there are no MongoDB version-related errors.
- Export and then import a knowledge base backup file. Check if the multi-file structure is retained.
- Upload a large-volume chromatography data file. Verify that the storage path and file integrity meet expectations.
The values provided are common starting points and should be measured against the reader's own samples.