Kubernetes observability using OpenTelemetry

Coralogix offers Kubernetes Observability using OpenTelemetry for comprehensive Kubernetes and application observability. Using our OpenTelemetry Chart, the integration enables you to simplify the collection of logs, metrics, and traces from the running application in your pods to the cluster-level components of your Kubernetes cluster.

Observability explained

Kubernetes observability

Kubernetes observability is essential for monitoring a Kubernetes cluster's health, performance, resource utilization, and workloads. It involves collecting and analyzing metrics, logs and traces from the cluster and underlying machines to ensure the stability and optimal operation of the cluster.

When managing and monitoring Kubernetes components, consider these critical areas:

• Cluster health: Monitoring the overall health of the Kubernetes cluster is crucial. This includes checking the status and availability of the master and worker nodes and the control plane components such as the API server, kube-proxy, and scheduler.

• Resource utilization: Observing the resource utilization of cluster nodes and individual pods is essential for identifying bottlenecks, optimizing resource allocation, and ensuring efficient utilization of cluster resources. Extracting metrics and metadata from the underlying components provides the CPU, memory consumption, system load, and file system activity.

• Networking: Monitoring Kubernetes networking is crucial for smooth pod and service communication. This involves observing network traffic, latency, and error rates to detect and troubleshoot connectivity issues, identify performance bottlenecks, and improve network configurations.

• Application performance: Observing the performance of applications running on Kubernetes is essential for delivering a reliable and responsive user experience.

• Logging and Tracing. Logging and tracing play a vital role in understanding the behaviour and troubleshooting of Kubernetes components and applications. By collecting and analysing logs and traces, you can gain insights into system events, diagnose issues, and perform root cause analysis. Implementing effective logging and tracing strategies is important to capture relevant information for observability purposes.

Application observability

Application observability focuses on monitoring and understanding the behavior of applications running on the Kubernetes cluster. It includes collecting and analyzing metrics, logs, and traces specific to the applications to gain insights into their performance and identify any issues or bottlenecks. This includes monitoring response times, throughput, error rates, and other application-specific metrics.

OpenTelemetry integration overview

Integrating OpenTelemetry with Kubernetes enables comprehensive Kubernetes and application observability. The OpenTelemetry Integration Chart is a solution that combines two dependent charts into a single Helm installation for Kubernetes clusters: the OpenTelemetry Agent and the OpenTelemetry Cluster Collector. Both are built on the OpenTelemetry Collector Helm Chart, but are configured for optimal performance while collecting different data sources from Kubernetes. Together, they simplify the collection of logs, metrics, and traces from the running application in pods to the cluster-level components of your Kubernetes cluster.

Depending on your needs, you can use both charts (default behavior) or decide to disable either one under the opentelemetry-agent or opentelemetry-cluster-collector sections in the values.yaml file.

OpenTelemetry Agent

The OpenTelemetry Agent simplifies the collection of logs, metrics, and traces from applications running in your Kubernetes cluster. It is configured to deploy as a daemonset and runs on every node in the cluster. The agent maps metadata - such as Kubernetes attributes, Kubelet metrics, and host data - to the collected telemetry. This is particularly beneficial for high-traffic clusters or when utilizing our APM capabilities.

The agent comes with several pre-configured processors and receivers:

• Coralogix Exporter. Coralogix Exporter automatically enriches telemetry data with Kubernetes attributes, using consistent ApplicationName and SubsystemName fields for better correlation across your telemetry signals.

• Kubernetes Attributes Processor. This processor enriches data with Kubernetes metadata, such as pod and deployment information.

• Kubernetes Log Collection. Enables native Kubernetes log collection with OpenTelemetry Collector, eliminating the need for multiple agents like Fluentd, Fluent Bit, or Filebeat.

• Host Metrics. native Linux monitor resource collection agent. No need to run Node exporter or vendor agents.

• Kubelet Metrics. Fetches running container metrics from the local Kubelet.

• OTLP Metrics. Sends application metrics using the OpenTelemetry protocol.

• Traces. Collects data in various formats such as Jaeger, OpenTelemetry Protocol, or Zipkin.

• Span Metrics. Converts optional traces into requests, duration, and error metrics using the span metrics processor.

• Zpages Extension. Investigate latency and error issues by navigating to the pod's localhost:55516 web server. Relevant routes are described in the OpenTelemetry documentation.

OpenTelemetry Cluster Collector

The OpenTelemetry Cluster Collector retrieves data from the cluster level, including Kubernetes events, cluster metrics, and additional Kubernetes-specific metrics. It enables you to gain insights into the health and performance of various objects within the cluster, such as deployments, nodes, and pods.

• Coralogix Exporter. Coralogix Exporter automatically enriches telemetry data with Kubernetes attributes, using consistent ApplicationName and SubsystemName fields for better correlation across your telemetry signals.

• Cluster Metrics Receiver. The Kubernetes Cluster receiver collects cluster-level metrics from the Kubernetes API server.

• Kubernetes Events Receiver - The Kubernetes Events receiver collects events from the Kubernetes API server. See Kubernetes Events for more information.

• Kubernetes Extra Metrics. This preset enables the collection of extra Kubernetes-related metrics, such as node information, pod status, or container I/O metrics.

• Integration Presets. This chart provides support to integrate with various applications (e.g. mysql) running on your cluster to monitor them out of the box.

Next steps

View our basic configuration instructions here.

Advanced configuration instructions can be found here.