Data Profile for This Category
The business data for in-feed ad creative testing comes primarily from ad delivery platforms, internal material libraries, user interaction data, and conversion data. Associated documents are updated daily or weekly, including creative scripts, material testing sheets, delivery data reports, and iteration checklists.
Data fields include material unique identifiers, creative combinations (images, videos, copywriting), delivery time windows, conversion costs, click-through rates, material fatigue tags, and other fields. No uniform fixed single data length exists. Support for multiple types of structured and unstructured storage is required. The size of some unstructured materials can reach multiple megabytes.
Constraints on Database and Operations Workflows
Ad platforms, material libraries, and other systems provide data sources. This requires support for synchronized writing and associated queries across multiple data sources.
Daily or weekly batch update cycles demand reasonable connection pool and batch processing configurations to avoid resource exhaustion. Structured testing sheets and iteration checklists need support for precise field filtering and version tracing. Unstructured creative materials need adaptation for large-field storage.
Data anomalies directly impact delivery performance. High database availability and complete operation log retention must be maintained to enable subsequent fault retrospective.
Configuration Guidelines
The following configurations are adapted for FastGPT v4.17.0.
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
MONGODB_URI | mongodb://[username:password@]host1:port1,host2:port2/database?authSource=admin&socketTimeoutMS=30000 milliseconds | Adapts to sharded storage for multi-source material libraries and delivery data, supports storage of large-field unstructured data |
PG_URL | postgresql://user:pass@host:port/dbname?sslmode=require&connect_timeout=30 seconds | Stores structured creative testing sheets, iteration checklists, and delivery metrics, ensures transactional data consistency |
DB_MAX_LINK | 20–30 connections | Balances write concurrency and query requests for daily batch data, prevents database connection exhaustion |
REDIS_URL | redis://:password@host:port/0?maxmemory-policy=allkeys-lru&timeout=600 seconds | Caches high-frequency queried creative combinations and real-time delivery metrics, reduces repeated query pressure on databases |
Three Common Misconfigurations
- Database connection requests are rejected, returning a 403 status code. The cause is that the public request IP of the deployment node was not added to the database whitelist.
- Multi-line query statements throw syntax errors when run in a MySQL environment. The cause is that session parameters allowing batch execution were not configured.
- Agents and knowledge bases do not appear after backup restoration. The cause is that only project files were restored, and associated business data in the database was not synchronized during restoration.
How to Verify Successful Configuration
- Run connectivity test commands on the database node to verify that the configured connection address can be accessed normally.
- Run a test script to insert a piece of creative testing data and execute a query to verify that database read and write functions work properly.
- Check Redis cache monitoring metrics to confirm that cache hit rates for high-frequency queries meet business expectations.
- Trigger a batch data synchronization task to verify that the database connection pool's concurrent processing capacity meets update requirements.
The values provided here are common starting points and should be evaluated against the samples of the respective deployment environment.
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.