Configuring the PGAI agent

Configuring the PGAI agent#

Note

TPA currently only supports using the PGAI Agent to integrate clusters with EDB Postgres AI Cloud Service. Using the agent to integrate with EDB Postgres AI Hybrid Manager, or for usage reporting, will be supported in a future release of TPA.

PGAI agent package version#

By default, TPA installs the latest available version of beacon-agent .

The version of the beacon-agent package that is installed can be specified by including beacon_agent_package_version: xxx under the cluster_vars section of the config.yml file.

cluster_vars:
    …
    beacon_agent_package_version: 1.56.2-1
    …

You may use any version specifier that apt or yum would accept.

If your version does not match, try appending a * wildcard. This is often necessary when the package version has an epoch qualifier like 2:... .

PGAI agent configuration#

The PGAI agent configuration contains two parameters which must be set per-cluster, the access key and the project id.

The access key is kept encrypted in the cluster directory and can be set or read using tpa’s store-password and show-password commands:

tpaexec store-password . beacon_agent_access_key

Password:

tpaexec show-password . beacon_agent_access_key

If the environment variable BEACON_AGENT_ACCESS_KEY is set when tpaexec provision is run, the access key is set from its value.

The project id is stored in config.yml under the beacon_agent_project_id key in cluster_vars . If the --beacon_agent_project_id argument is passed to tpaexec configure then its value is written to config.yml appropriately.

Installing the PGAI agent#

TPA installs the PGAI agent from EDB’s repositories and creates an operating system user called beacon and a database user called beacon . A configuration file for the agent is written to .beacon/beacon_agent.yaml in the beacon user’s home directory.

Running the PGAI agent#

TPA installs a systemd service unit file to start the agent at boot-time, running as the beacon user.

Contents