How to set up Prometheus monitoring for your services

Yuri, identified in the transcript as an SRE at Google, demonstrates one Python service instrumented for Google Managed Service for Prometheus. The application exposes request, failure, and latency metrics; managed collection scrapes them; and a PromQL query verifies ingestion before the data is used in charts or alerts ([00:00:00]-[00:03:11]).

Key Points Covered

• Application signals complement infrastructure metrics: The sample adds counters and a latency distribution for service behavior that infrastructure telemetry alone cannot explain ([00:00:00]-[00:01:03]).
• Collection starts at an explicit endpoint: The application exposes a Prometheus /metrics endpoint, while a PodMonitoring resource selects pods and defines how the collector should scrape it ([00:01:03]-[00:03:11]).
• Verify the path end to end: The walkthrough checks collector self-metrics and then queries the application metric before relying on Metrics Explorer, dashboards, or alerts ([00:02:06]-[00:03:11]).
• The example is deliberately bounded: It does not address cardinality, retention, cost, authentication, or alert thresholds, and the auto-captioned port number may be wrong.

This is a 2022 walkthrough, not current configuration guidance. Verify today's commands, managed-collection behavior, resource schema, metric conventions, and console UI before use.

Full video: https://www.youtube.com/watch?v=qyfOE_78nT0(opens in a new tab)