Collect CloudWatch Database Insights metrics
The database_insights source collects metrics and logs from self-managed
databases for CloudWatch Database Insights. This source can contain a postgresql
array. Each entry in the array can include the following fields:
-
endpoint– Required. The host and port that the agent connects to, for examplelocalhost:5432. -
instance_name– Required. The display name of the instance in Database Insights. -
username– Required. The database user that the agent uses for monitoring. -
password_file– Required. The path to a libpq pgpass file that contains the monitoring user's password. -
ca_file– Optional. The path to a CA certificate file for TLS connections to the database. -
per_resource_collection_interval– Optional. How often, in seconds, to collect per-table and per-index metrics. The default is 60. -
logs– Optional. Specifies the database log files to collect, with afile_pathfield.
In the monitored PostgreSQL database, create a monitoring role that the agent uses, grant
it the pg_monitor role, and enable the pg_stat_statements extension.
Store the monitoring role's password in a libpq pgpass file and point password_file
at it. For the complete setup procedure, including how to create the monitoring user, the
pgpass file, and the full agent configuration, see Monitoring Self-Managed PostgreSQL.
The following example opentelemetry section collects Database Insights metrics
from a self-managed PostgreSQL database that runs on the same host as the agent.
{ "opentelemetry": { "collect": { "database_insights": { "postgresql": [ { "endpoint": "localhost:5432", "instance_name": "example-postgres", "username": "cwagent", "password_file": "/opt/aws/amazon-cloudwatch-agent/etc/cwagent-pgpass", "per_resource_collection_interval": 60 } ] } } } }
The agent publishes these as OpenTelemetry metrics to the CloudWatch OTLP endpoint, which you view in CloudWatch Database Insights and query with PromQL.