Origin Metrics is an OpenShift component designed to collect and manage container, pod, and node metrics across entire OpenShift clusters. Written primarily in Shell, the repository provides infrastructure for comprehensive cluster monitoring and performance visibility. The project is classified within the metrics collection, monitoring, and Kubernetes domains, with specific focus on Prometheus integration patterns, cluster performance analysis, resource usage tracking, time series data management, dashboard visualization, and alerting capabilities.
The system architecture consists of four primary components working in concert. Heapster serves as the metrics collection agent, gathering system-level metrics such as CPU, memory, and network statistics from across the OpenShift cluster. It retrieves metadata from the cluster master API and collects individual metrics from the stats endpoint exposed on each OpenShift node. Heapster then forwards these metrics to Hawkular Metrics and exposes a REST endpoint used by the horizontal pod autoscaler for scaling decisions. Hawkular Metrics functions as the metric storage engine, providing a JSON-based REST interface for creating, accessing, and managing historically stored metrics. It receives metrics from Heapster and exposes them to the OpenShift Console and third-party systems. Cassandra serves as the underlying datastore for Hawkular Metrics, persisting all incoming metric data. The Hawkular OpenShift Agent, currently in Tech Preview, enables collection of application-level metrics directly from pods, running as a daemon set across the cluster to gather metrics that individual pods expose.
Installation methodology has evolved significantly within this project. The repository documentation explicitly notes that metrics installation has transitioned from using a deployer pod approach to an Ansible-based installation method, with the older deployer pod approach now deprecated. The project maintains backward compatibility documentation for the legacy deployment method while directing users toward the current Ansible-based installation process documented in the official OpenShift documentation.
The repository provides multiple access patterns for metrics data. Users can access metrics through the OpenShift Console, query the Hawkular Metrics REST API directly for raw metrics data and export capabilities, or utilize the Hawkular Metrics Python client which is pre-installed with admin privileges in the Hawkular Metrics pod. Heapster is accessible exclusively through the API Proxy and requires cluster-admin privileges. The project includes comprehensive deployment instructions for the Hawkular OpenShift Agent, requiring ConfigMap creation, template processing, and service account permission configuration.
The repository includes operational guidance for cleanup and removal of deployed components, with specific commands for deleting metrics-related resources while preserving persistent volume claims containing historical data. Docker images for all Origin Metrics components are available on Docker Hub, eliminating the need for most users to build images directly. The project maintains documentation covering build instructions, troubleshooting guidance, known issues, and direct Hawkular Metrics API usage. The metrics system requires a properly installed and configured OpenShift cluster as a prerequisite, with common installation and configuration problems checked during deployment to provide diagnostic error messages.
