Kubernetes metrics

Note: This page describes the newly-released version 1.x of the Kubernetes integration. If you're running the Automated dashboard-based version, see our Kubernetes 0.x documentation.

The AppSignal for Kubernetes integration extracts metrics from Kubernetes clusters.

After adding the AppSignal Kubernetes deployment to your cluster, a Cluster Metrics page appears in AppSignal, showing per-node and per-pod metrics.

Installation

To start reporting Kubernetes cluster metrics to AppSignal:

  1. Store your AppSignal API key in a secret
  2. Add AppSignal deployment to your cluster

1. Store your AppSignal API key in a secret

In a Kubernetes cluster, set up your AppSignal API key (find your App-specific API key in App settings) by creating a secret:

Shell
kubectl create secret generic appsignal --from-literal=api-key=<APPSIGNAL_API_KEY>

2. Add AppSignal deployment to your cluster

Once you've created the AppSignal API key secret, add the AppSignal deployment to your cluster:

Shell
kubectl apply -f https://raw.githubusercontent.com/appsignal/appsignal-kubernetes/main/deployment.yaml

Once configured, AppSignal for Kubernetes will start reporting Kubernetes metrics automatically.

Cluster Metrics

In AppSignal, the Host Metrics page will now be replaced with a Cluster Metrics page to show information about pods and nodes in your cluster. You will be able to see graphs representing useful metrics over time, such as:

  • CPU Usage: CPU usage stats per node or pod
  • Memory Usage: memory usage statistics, comparing the used memory to what's available per node or pod
  • Swap Usage: swap usage statistics, comparing the used swap to what's available per node or pod
  • Disk Usage: disk usage statistics, comparing the used disk space to what's available per node
  • Network usage: network usage, split by received and transmitted bytes, per node

Support

If you are experiencing any difficulties configuring your application's Kubernetes metrics, contact support, and we'll help you get your metrics flowing.