o11y's Introduction

RHTAP Observability

This repository contains Prometheus recording rule files and alert rule files for observability use, as well as unit tests for these rules.

Recording Rules

appstudio_container_network_transmit_bytes_total collects the network egress of containers and adds the label_pipelines_appstudio_openshift_io_type label to our metric.
appstudio_container_resource_limits collects all the resource limits (cpu and memory) for all containers and init_containers, and adds the label_pipelines_appstudio_openshift_io_type label to the new metric
appstudio_container_resource_minutes_gauge uses the resource limit configured on the container, as captured by appstudio_container_resource_limits, and transforms it to a fixed time window gauge representing the resource-limit per minute multiplied by the period the container was alive in that time window.

Examples (assuming 1-minute time window)

A container lived for the entire minute and had a limit of 0.5 CPU cores --> 0.5 CPU minutes.
A container lived for 30 seconds within the measured time frame and had a limit of 2 CPU cores --> 1 CPU minutes. This metric is intended to then be summed over the time frame the metric consumer wants to measure for.

Limitations

When a pod contains one or more init_containers that transmit data, only the first init_container will be taken into account while the other init_containers and containers are ignored from the metric appstudio_container_network_transmit_bytes_total

Recommend Projects