Monitoring your Kubernetes cluster is essential for ensuring performance, stability, and reliability. K8s Infra, the SigNoz Kubernetes collection agent, provides a comprehensive solution for collecting and analyzing metrics, traces, logs, and Kubernetes events from your entire Kubernetes environment by using OpenTelemetry Collectors.
Install it with the k8s-infra Helm chart, or with Foundry through Helm or Kustomize; each method installs the two pre-configured OpenTelemetry Collectors described below across your Kubernetes cluster.
K8s Infra primarily performs the following:
- Tails and parses logs generated by containers in the Kubernetes cluster and sends them to SigNoz.
- Gathers Traces from your Kubernetes Cluster generated by containers and sends them to SigNoz
- Collects kubelet metrics and host metrics from each node in the Kubernetes cluster.
- Gathers cluster-level metrics from the Kubernetes API server.
- Acts as a gateway to send any incoming OTLP telemetry data to the SigNoz OTel Collector.
How it Works: The otelAgent and otelDeployment
To export Kubernetes signals, you can enable different receivers in the OpenTelemetry collector, which will send signals about your Kubernetes infrastructure to SigNoz. These OpenTelemetry collectors will act as agents that transmit signals about Kubernetes to SigNoz. K8s Infra deploys two OpenTelemetry Collectors to ensure complete observability of your Kubernetes cluster.
OpenTelemetry Collector as an Agent: otelAgent
otelAgent is deployed as a DaemonSet, meaning an instance of the otelAgent runs on every node in your cluster. Its main task is to collect node-level and pod-level data, including:
- Host Metrics: CPU usage, memory consumption, disk I/O, and network statistics from each node.
- Kubelet Metrics: Data from the Kubelet API on each node, such as pod and container resource utilization.
- Container Logs: Tailing and parsing logs generated by the containers running on the node.
- Collect Traces: Traces from the application will also be collected generated by containers on node.
OpenTelemetry Collector as a Deployment: otelDeployment
The otelDeployment collector is set up to gather cluster metrics and events by default.
- Cluster-level Metrics: Information about Kubernetes objects such as Deployments, StatefulSets, DaemonSets, and Jobs.
- Kubernetes Events: Significant events within the cluster, useful for troubleshooting and auditing.
- Custom Metrics: It can be configured to collect metrics from other services in the cluster, like Redis or PostgreSQL.
- Prometheus Scraping: Metrics can also be scraped using pod annotations.
The otelAgent collector is deployed as a daemonset, which can lead to duplicate metrics if the same component runs on multiple nodes and metrics are collected from each node. The otelDeployment collector resolves this issue by collecting metrics at the cluster level instead of from individual nodes.
