Skip to main content

Production observability and cost reporting were brought up to date

ยท 2 min read
Atlas Infra

The Atlas production root now has the same observability reporting shape documented for the active environment: New Relic AWS pull integration, service dashboards, RDS/MSK dashboards, and ClickHouse Cloud dashboard outputs are available from Terraform outputs.

This update aligns the handbook and operator-facing report with the current production outputs:

  • production exposes the public ALB, ECS service names, ECR repositories, RDS endpoints, MSK Connect sink, and New Relic dashboard permalinks
  • New Relic AWS pull integration is enabled in both committed roots
  • the ClickHouse Cloud Prometheus agent resources and New Relic dashboard are enabled in both roots
  • the Prometheus collector desired count remains a separate control and is currently 0 in committed values
  • production keeps the dedicated AWS/SNS Slack alarm delivery path disabled until a real webhook is injected outside Git
  • staging keeps Slack delivery enabled; DevOps Agent investigation wiring remains only a test path

The cost section now separates AWS forecasts from ClickHouse Cloud usage:

  • AWS staging forecast for May 2026: $1,547.57
  • AWS production forecast for May 2026: $675.80
  • ClickHouse Cloud billable usage for Apr 3-May 3, 2026: $719.14

The ClickHouse usage statement is approximate until the billing period finalizes, so it should be treated as a current billing signal rather than a closed invoice.

The related documentation updates also corrected stale references to staging-only ClickHouse observability, production-disabled New Relic AWS pull integration, disabled Container Insights, and outdated MSK sizing notes.