Logs, metrics, and usage

Autodisc separates build output, runtime logs, service metrics, and account usage because they answer different questions.

Build output

Open Service → Deployments, choose a deployment, and select its build or runtime output. Use this for dependency installation, compilation, image pull, and release failures.

Runtime logs

Open Service → Logs for application output. The live view follows the running service; retained periods are available when log history is enabled.

Good application logs:

• identify the service and operation;
• include stable request or job correlation IDs;
• explain retries and terminal failures; and
• never print credentials, authorization headers, database URLs, or private tokens.

Metrics

Open Service → Metrics for CPU and memory behavior. Compare usage with the configured limit:

• sustained memory growth can indicate a leak or insufficient memory;
• CPU saturation can indicate expensive work or insufficient compute;
• an immediate drop to zero can indicate a crash or stopped container.

Project usage

Open Project Settings → Usage for environment services and account-plan meters such as GB-hours and build minutes.

GB-hours measure memory allocation over runtime. A 512 MB service running for 24 hours uses 12 GB-hours.

Incident workflow

1. Confirm the environment and affected service.
2. Check deployment health and recent changes.
3. Inspect build output if the service never started.
4. Inspect runtime logs if it started and failed.
5. Compare metrics around the failure time.
6. Check variables, managed-resource health, and networking.
7. Roll back or request an agent repair Change Request when the evidence points to a source change.