What the data for this category looks like
The data sources for this category typically include user behavior logs, push systems, experiment platforms, and CRM systems. Data update frequency is daily or weekly. Corresponding daily produced documents include growth experiment plans, user segmentation tables, outreach copy, and retention analysis reports. Document structure is usually divided by experiment cycle, user segmentation dimension, and outreach channel. Fields include user ID, outreach count, retention cycle, and experiment group ID. There are no mandatory additional units; record formats are matched to business scenarios only.
What constraints do these characteristics impose on deployment and upgrades
Because data sources involve multiple internal systems, deployments must configure intranet-accessible database, cache, and model API connection parameters to avoid cross-network access permission issues. The daily or weekly data update frequency requires the knowledge base sync cycle to match business rhythms, preventing a disconnect between documents and real-time data. Daily documents have diverse types, so storage paths must be configured by category in the knowledge base, such as experiment plans and user segmentation tables, to facilitate retrieval by operations personnel. Additionally, internal ticket and on-call scenarios involve sensitive user data, so access permissions must be strictly controlled to prevent leakage of operational secrets. If a vector storage knowledge base is used, upgrading the PGVector plugin must be compatible with existing data structures to avoid impacting historical experiment data queries.
How to set configurations
The following configurations are based on the open-source deployment process for FastGPT v4.17.0.
| Configuration Item | Recommended Value | Rationale |
|---|---|---|
MONGODB_URI | mongodb://operations system account:password@intranet IP:27017/fastgpt | Stores user segmentation tables, experiment logs, and other operational data; requires intranet access to ensure data security |
PG_URL | postgresql://operations and maintenance account:password@intranet IP:5432/fastgpt | Stores internal ticket records and knowledge base metadata; adapted for intranet deployment environments |
REDIS_URL | redis://cache account:password@intranet IP:6379/0 | Caches user sessions and temporary outreach copy data to reduce database concurrent pressure |
FE_DOMAIN | http://intranet fastgpt service domain:3000 | Validates domain legitimacy when generating login-free links; adapted for intranet access scenarios |
DB_MAX_LINK | 20–30 connections | Matches daily or weekly bulk data sync requests from operations systems to avoid exhausting database connections |
ROOT_KEY | Randomly generated string of 32 or more characters | Controls administrator permissions for internal ticket Q&A to prevent unauthorized access to sensitive operational data |
Three common mistakes
- Phenomenon: After a user using a login-free link deletes conversation content, no corresponding record appears in backend logs. Cause: The log association rule for
FILE_TOKEN_KEYwas not configured correctly, causing the binding relationship between conversation ID and log records to be lost. - Phenomenon: Calling the internal DeepSeek API fails after intranet deployment, returning a connection timeout error. Cause: A public network API address was used incorrectly, and no intranet-accessible model API endpoint was configured.
- Phenomenon: An error about the quantity limit is prompted when adding new operation-related knowledge bases, making it impossible to continue storing documents such as user segmentation tables and growth experiment plans. Cause: The corresponding quota parameter was not adjusted, and the default 30 knowledge base limit of the open-source version was retained.
How to confirm configurations are correct
- Access the configured
FE_DOMAIN, generate a login-free link, delete the conversation, and check backend logs to confirm that log records with the corresponding conversation ID exist. - Attempt to add more than 30 knowledge base documents, confirm that submission and storage can proceed normally, with no quota limit prompts.
- Use the command line to connect to
PG_URLin the intranet environment, executeSELECT extversion FROM pg_extension WHERE extname = 'vector';, and confirm that the PGVector version meets expected requirements. - Use a Redis client to connect to
REDIS_URLin the intranet environment, execute theKEYS *command, and confirm that cached operational user session data can be read normally.
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.