Infrastructure Monitoring

Osuite monitors your infrastructure with the OpenTelemetry Collector — no proprietary agents, no vendor-specific exporters. Host, container, cluster, and service metrics arrive over OTLP like every other signal, so they share resource attributes with your traces and logs. From a degraded service in APM you can move to the host or pod underneath it, and the Investigation Agent can tell whether an incident is the application’s fault or the machine’s.

Each guide below is a complete, copy-paste Collector configuration for one source.

Hosts

Collect CPU, memory, disk, filesystem, network, and load metrics with the Collector’s hostmetrics receiver, running as a service on the host.

• Linux host monitoring — hostmetrics receiver, Collector as a systemd service
• Windows host monitoring — hostmetrics plus Windows performance counters, Collector as a Windows service

Containers

• Docker container metrics — per-container CPU, memory, network, and block I/O with the docker_stats receiver. For container logs, see Collect Docker logs.

Kubernetes

The osuite-k8s Helm chart deploys the Collector cluster-wide and as a daemonset, collecting cluster-state metrics, node and pod resource metrics, container logs, and cluster events.

• Kubernetes monitoring — install the chart and see what it collects
• Cluster events — collect and search Kubernetes events
• Container logs — per-namespace and per-severity log collection
• Manual collector setup — the raw Collector configuration, without the chart

Integrations

Monitor databases, message queues, and web servers with the Collector’s per-service receivers. Each integration guide follows the same shape: install, configure, validate, and a table of the metrics collected.

• Databases — PostgreSQL, MySQL, Redis, MongoDB, Elasticsearch
• Message queues — Kafka, RabbitMQ
• Web servers and proxies — nginx, Apache, HAProxy
• Cloud — AWS, GCP, Azure

Correlated with your applications

Infrastructure metrics are not a separate silo. They carry the same resource attributes as your APM, logs, and traces, so during an incident you can move between a degraded service and the host or pod metrics beneath it. The Investigation Agent correlates the two automatically when it builds an incident timeline.

Next steps

• Dashboards — browse the pre-built Kubernetes dashboards, or build your own
• Alerts — alert on resource thresholds before utilization becomes an incident
• APM & Tracing — correlate infrastructure health with service performance