Metrics service configuration variables

Review configuration variables to customize metrics service deployment, database connections, security settings, and operational parameters.

Inventory variables reference

Required variables for automationmetrics group

The following variables must be defined when deploying metrics service with the containerized installer:

Variable Default Required/Optional Description
automationmetrics_pg_host (none) Required PostgreSQL host for the metrics database
automationmetrics_pg_password (none) Required* Password for the metrics database user. *Not required whenautomationmetrics_pg_cert_auth: true is set.
automationmetrics_controller_read_pg_host (none) Required* PostgreSQL host for the controller database
automationmetrics_controller_read_pg_password (none) Required* Password for the read-only controller database user. *Not required whenautomationmetrics_pg_cert_auth: true is set.

Optional variables with defaults

Variable Default Description
automationmetrics_pg_port 5432 PostgreSQL port
automationmetrics_pg_database metrics_service Database name
automationmetrics_pg_username metrics_service Database user
automationmetrics_api_port 8006 Port the metrics service API listens on (Gunicorn)
automationmetrics_nginx_http_port 8087 nginx HTTP port (redirects to HTTPS)
automationmetrics_nginx_https_port 8450 nginx HTTPS port
automationmetrics_pg_cert_auth false Use certificate authentication instead of passwords
Note

Note on certificate authentication: When automationmetrics_pg_cert_auth: true, password variables (automationmetrics_pg_password and automationmetrics_controller_read_pg_password) are not required. The service uses PostgreSQL certificate-based authentication instead.

Indirect node audit data retention
Copy link

Collection alone is not sufficient: the source audit records must still exist in the controller database when a collection runs. The controller retains indirect node audit records in the main_indirectmanagednodeaudit table for a limited period, then purges older records. If audit records are purged before they are collected, that data is lost and cannot be recovered. The controller retention period does not apply to metrics service data. The controller stores raw indirect node audit records (one per host per job), which grow large and are purged after 7 days by default. Metrics service stores only aggregated daily counts by collection and module, which are small and can be retained for much longer.

Variable

Default

Level

Description

INDIRECT_HOST_AUDIT_RECORD_MAX_AGE_DAYS

7

Automation controller setting

Number of days the automation controller retains indirect node audit records before purging them. This is an automation controller setting, not a metrics service automationmetrics_* inventory variable. It is set as a default and can be overridden with an environment variable or settings file. It is not exposed as a controller UI or API configuration setting.
Note
The controller purges indirect node audit records after the retention period (7 days by default). Any process that reads these records, such as the metrics service daily collection and the metrics utility gather, must run within the retention window, or purged records are lost permanently. If you run the metrics utility gather less frequently than the retention period (for example, monthly), increase INDIRECT_HOST_AUDIT_RECORD_MAX_AGE_DAYS to cover the full interval between gathers.
Automation controller has four indirect-node-related settings. Only one is exposed through a supported customer interface. The other three are Django settings, changeable only through an environment variable or settings file:

Variable

Default

Customer-togglable

Description

INDIRECT_NODE_COUNTING_ENABLED

True

Yes. Settings > System UI, the /api/v2/settings/system/ API, or awx-manage set_indirect_node_counting

Controls whether the controller records indirect node audit data at all.

INDIRECT_HOST_AUDIT_RECORD_MAX_AGE_DAYS

7

No. Settings file or environment variable only

Retention window for raw audit records, described above.

INDIRECT_HOST_QUERY_FALLBACK_MINUTES

60

No. Settings file or environment variable only

How long the controller waits before the hourly fallback job picks up a job whose indirect node audit record was not recorded by the immediate path.

INDIRECT_HOST_QUERY_FALLBACK_GIVEUP_DAYS

3

No. Settings file or environment variable only

How long the controller keeps retrying a job whose events failed to process before giving up.
Note

If a job's events fail to process for longer than INDIRECT_HOST_QUERY_FALLBACK_GIVEUP_DAYS (3 days by default), the controller gives up and never creates an audit record for that job. This loss is permanent and independent of the 7-day retention window. Increasing INDIRECT_HOST_AUDIT_RECORD_MAX_AGE_DAYS does not recover it, because there was never a record to purge.

Events collector settings

The events collector (main_jobevent_service) processes job event data to extract collection, module, and role usage statistics. These variables are configurable as Django settings or environment variables without a code change. Use them to control collection volume on large deployments.

Variable Default Description
JOBEVENT_ROW_LIMIT 200,000 Maximum number of job event rows to process per hourly collection cycle. Set to 0 for unlimited processing.
JOBEVENT_JOB_LIMIT 1,000 Maximum number of jobs to process for event data per hourly collection cycle. Set to 0 for unlimited processing.
Warning
Setting JOBEVENT_ROW_LIMIT or JOBEVENT_JOB_LIMIT to 0 removes the limit entirely, causing the collector to process all available events. On large deployments with high job volumes, this can significantly increase collection time and database load. To effectively disable event collection for performance reasons, set both values to 1.