Azure Monitor SLIs on Managed Prometheus: the bits the docs don't tell you

This article details the author's experience creating Azure Monitor SLIs (Service Level Indicators) against Managed Prometheus metrics on AKS, using infrastructure as code. It highlights several undocumented pitfalls, most notably that the metric namespace for Managed Prometheus is 'customdefault' not 'prometheus', which causes confusing errors. The post covers the SLI resource structure (extension resource on a service group), the API version (2025-03-01-preview), and practical tips for applying SLIs via ARM calls. It's a technical guide for DevOps engineers and SREs working with Azure Monitor and Prometheus.