What data for this category looks like
Text annotation project data primarily comes from customer corpora, annotation results from annotation platforms, custom vocabularies, and quality inspection sampling results. Associated daily documents include annotation specifications, pending annotation sample packages, quality inspection reports, and issue feedback forms. Data updates follow a daily or per-batch annotation schedule. Each individual data entry includes fields such as text content, annotation tags, annotator identifier, quality inspection result, and annotation time spent. Fields and units are adjusted based on specific annotation tasks, with no unified fixed format.
What constraints do these characteristics impose on the database and operations link
The data characteristics of text annotation projects create multiple constraints for database and operations workflows. Integrate data from multiple sources: customer corpora, annotation platform results, custom vocabularies, and quality inspection data. The system must support multiple data formats and access methods. Adhere to daily or per-batch update schedules. The database must support batch writes and incremental synchronization to avoid performance loss from full re-imports. Support multi-role business processes. Configure fine-grained database access permissions to distinguish operational scopes for annotators, quality inspectors, and administrators. Address the semantic nature of text data. The storage layer must support semantic association retrieval, while ensuring consistency between annotation and quality inspection data to prevent issues caused by misalignment.
How to configure settings
| Configuration Item | Recommended Setting | Rationale for This Setting |
|---|---|---|
MONGODB_URI | mongodb://<user>:<pass>@<host>:27017/text_label?authSource=admin | Stores unstructured text, annotation tags, and other data, adapting to the flexible storage needs of text annotation |
PG_URL | postgresql://<user>:<pass>@<host>:5432/label_ops?sslmode=require | Stores structured business data such as annotation specifications and quality inspection reports, supporting transactional operations |
DB_MAX_LINK | 20–30 connections | Matches the concurrent connection requirements of daily batch annotation tasks, preventing connection pool resource exhaustion |
REDIS_URL | redis://:<pass>@<host>:6379/1 | Caches frequently called vocabularies and annotation rules to reduce database query latency |
| Database connection timeout threshold | 30 seconds | Adapts to the wait time for multi-data source integration, preventing slow external data source responses from blocking tasks |
Three common configuration mistakes
- Symptom: MongoDB connection status query returns failure, logs show authentication error. Cause: The correct authentication source or account password is not configured in
MONGODB_URI, preventing normal database access. - Symptom: No corresponding option appears when attempting to configure an Oracle database connection, making access impossible. Cause: FastGPT v4.17.0 currently only supports configuring
MONGODB_URIandPG_URLdatabase connection parameters, and access configuration for Oracle databases is not enabled. - Symptom: An error occurs when connecting to a MSSQL database in an orchestration workflow, with a connection timeout prompt, but local visualization tools can connect normally. Cause: The network policy of the FastGPT runtime environment restricts outbound port access, or the database address configured in the orchestration workflow does not match the local test address, preventing connection establishment.
How to confirm configuration is complete
- Run the built-in database connectivity test tool to verify that databases corresponding to
MONGODB_URIandPG_URLcan establish normal connections and perform read/write operations. - View the system connection pool monitoring panel to confirm that the current active database connection count does not exceed the configured maximum connection limit.
- Submit a batch of test annotation data to check whether data can be synchronized to the corresponding database according to the preset update schedule.
- Access the annotation rule query interface to confirm that frequently accessed vocabularies and annotation specifications are read from the cache without directly querying the database.
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.