Kubernetes

Kubernetes, also called K8s, is an open-source system for automating deployment, scaling, and managing containerized applications. K8s groups containers that make up an application into logical units for management and discovery.

Observe provides a manifest to collect telemetry within a Kubernetes cluster. By default, Observe gathers all events, logs, and metrics within a Kubernetes cluster using open-source collectors.

When you install Kubernetes using the Observe Kubernetes app, the app installs the following datasets, dashboards, and monitor templates by default:

Note

You can find more details about Kubernetes in the Kubernetes documentation.

Event Datasets

• kubernetes/API Update

• kubernetes/CAdvisor Derived Metrics

• kubernetes/CAdvisor Metrics

• kubernetes/Container Logs

• kubernetes/Container Spec

• kubernetes/Event

• kubernetes/Events

• kubernetes/Metrics

• kubernetes/Pod Metrics

• kubernetes/Pod Update

Resource Datasets

• kubernetes/Cluster

• kubernetes/Condition

• kubernetes/Config Map

• kubernetes/Container

• kubernetes/CronJob

• kubernetes/DaemonSet

• kubernetes/Deployment

• kubernetes/Endpoint

• kubernetes/Endpoints

• kubernetes/Environment Variables

• kubernetes/Image

• kubernetes/Image Repository

• kubernetes/Ingress

• kubernetes/Job

• kubernetes/Namespace

• kubernetes/Node

• kubernetes/Node Condition

• kubernetes/Object

• kubernetes/Persistent Volume

• kubernetes/Persistent Volume Claim

• kubernetes/Physical Container

• kubernetes/Pod

• kubernetes/Pod Condition

• kubernetes/ReplicaSet

• kubernetes/Secret

• kubernetes/Service

• kubernetes/StatefulSet

• kubernetes/Volume

• kubernetes/Volume Mount

Dashboards

• kubernetes/CAdvisor Metrics by Cluster

• kubernetes/CAdvisor Metrics by Container

• kubernetes/CAdvisor Metrics by Namespace

• kubernetes/CAdvisor Metrics by Node

• kubernetes/CAdvisor Metrics by Pod

• kubernetes/Kubernetes Home

• kubernetes/Go Collector Metrics

• kubernetes/Observe Metrics

Monitor Templates

• kubernetes/Container

• kubernetes/Container OOMKilled

• kubernetes/CronJob

• kubernetes/DaemonSet

• kubernetes/Deployment

• kubernetes/Endpoint

• kubernetes/Job

• kubernetes/Misc

• kubernetes/Node

• kubernetes/Node Memory Pressure

• kubernetes/Node Network Unavailable

• kubernetes/Node PID Pressure

• kubernetes/Node Ready

• kubernetes/Pod

• kubernetes/Pod Containers Ready

• kubernetes/Pod Initialized

• kubernetes/Pod Ready

• kubernetes/Pod Scheduled

• kubernetes/ReplicaSet

Viewing Kubernetes Data in Observe Dashboards

When you install Kubernetes on Observe, the Kubernetes app installs relevant dashboards by default.

kubernetes/CAdvisor Metrics by Cluster

View your cluster-wide resource usage and performance on this dashboard. Use the Group By menu to determine how you want to group the information. Select from the following groups:

• Cluster

• Container

• Namespace

• Node

• Pod

Figure 1 - Select from the Group By menu

Figure 2 - kubernetes/CAdvisor Metrics by Cluster Dashboard

The dashboard displays the following information about your Kubernetes clusters:

• Summary

  • CPU Usage

  • Memory Usage

• Node Utilization

  The following utilization metrics are computed as the sum of CPU and memory across all containers running on a given node. This value may underestimate the actual utilization by not accounting for processes outside of Kubernetes.

  • Total Container CPU Utilization Per Node

  • Total Container Memory Utilization Per Node

• Limits

  Resource utilization in Kubernetes is calculated relative to container limits. Containers that exceed their CPU limit in Kubernetes are throttled. Containers that exceed their memory limit are killed.

  • Max Container CPU Utilization

  • Max Container Memory Utilization

• Network

  • Inbound Throughput

  • Outbound Throughput

• Container Filesystem

  • Total Reads

  • Total Writes

  • File Descriptors

For the Kubernetes app installation instructions, see Installing the Kubernetes App.