Scrape Astra DB Serverless metrics in exposition format

This Astra DB Serverless feature is currently in public preview. Development is ongoing, and the features and functionality are subject to change. Astra DB Serverless, and the use of such, is subject to the DataStax Preview Terms.

You can use compatible third-party observability tools and services to scrape Astra DB Serverless metrics in Prometheus exposition format. You can use any tool or service that is compatible with Prometheus exposition format, such as Prometheus, Amazon CloudWatch, Datadog, Victoria Metrics, and more. The following steps use Prometheus for example purposes.

Whereas exporting metrics is push-based, this approach is pull-based. Your security policies might require you to use a pull-based configuration for metrics ingestion.

Prerequisites

Scraping metrics is a premium feature that requires a paid subscription plan.

Get credentials and IDs

Create an application token with a role that has Manage Metrics permission.

If you plan to scrape multiple databases, you can customize each token’s scope. For example, you can create separate tokens for each database, or you can create one token with the Organization Administrator role.
Get the database ID for each Astra DB Serverless database that you want to scrape.
If your Astra organization has PCU groups, and you want to scape PCU group metrics, get the uuid for each PCU group that you want to scrape.

To get a list of PCU groups in your Astra organization, use POST /v2/pcus/actions/get:
```
curl -sS -L -X POST "https://api.astra.datastax.com/v2/pcus/actions/get" \
--header "Authorization: Bearer APPLICATION_TOKEN" \
--header "Content-Type: application/json" \
--data '{}'
```
The provided APPLICATION_TOKEN must have the Organization Administrator or Billing Administrator role.

Configure Prometheus to scrape metrics

In your Prometheus configuration YAML file, in the scrape_configs section, add a job for each database that you want to scrape:

scrape_configs:
  - job_name: "prometheus-FIRST_DATABASE_ID"
    scrape_interval: 30s
    metrics_path: /v1/databases/FIRST_DATABASE_ID/metrics
    scheme: "https"
    authorization:
      type: "Astra-Token"
      credentials: "APPLICATION_TOKEN"
    static_configs:
      - targets: ["metrics.astra.datastax.com"]

  - job_name: "prometheus-SECOND_DATABASE_ID"
    scrape_interval: 30s
    metrics_path: /v1/databases/SECOND_DATABASE_ID/metrics
    scheme: "https"
    authorization:
      type: "Astra-Token"
      credentials: "APPLICATION_TOKEN"
    static_configs:
      - targets: ["metrics.astra.datastax.com"]

For databases in PCU groups, add a job for each PCU group that you want to scape:

scrape_configs:
  - job_name: "prometheus-PCU_GROUP_ID"
    scrape_interval: 30s
    metrics_path: /v1/pcugroup/PCU_GROUP_ID/metrics
    scheme: "https"
    authorization:
      type: "Astra-Token"
      credentials: "APPLICATION_TOKEN"
    static_configs:
      - targets: ["metrics.astra.datastax.com"]

Set the global or job scrape interval to 30s. This is the recommended interval to avoid dropping histogram buckets and stay in sync with the Astra Portal dashboard.

# Set an interval for all jobs in the global section.
global:
  scrape_interval: 30s

# Set intervals for individual jobs in the scrape_configs section.
scrape_configs:
  - job_name: "prometheus-DATABASE_ID"
    scrape_interval: 30s

Reload your Prometheus configuration.
If required, configure your observability service to receive the Astra DB metrics from your Prometheus endpoint.

Scrape Astra DB Serverless metrics in exposition format

Prerequisites

Get credentials and IDs

Configure Prometheus to scrape metrics

See also

Was this helpful?

Give Feedback