Top
Enterprise Postgres 18 Knowledge DataManagement FeatureUser's Guide

3.3.1 Configuring Embedded Providers (for Workers)

Before defining a vectorization, you must configure the worker process that will perform the vectorization to access the embedded provider.

Example) Settings when using Ollama as an embedded provider

rag_database=> SELECT pgx_vectorizer.set_worker_setting('ollama', 'OLLAMA_BASE_URL', 'http://your.ollama.server:11434');

Example) Settings when using OpenAI as an embedded provider

rag_database=> SELECT pgx_vectorizer.set_worker_setting('openai', 'OPENAI_API_KEY', 'your api key');

Information

Automatic vectorization is performed by worker processes that are regularly initiated. Additionally, the worker processes are initiated by Fujitsu Enterprise Postgres.

Point

In this configuration, credentials for the worker processes to access the embedded provider are stored in an area accessible by Fujitsu Enterprise Postgres. Because the configuration information for the embedded provider, including the API key, is stored in the table as plain text, place the table in an encrypted tablespace and protect it with the Fujitsu Enterprise Postgres transparent data encryption feature. For instructions, refer to "3.2.5 Credentials Protection".