https://github.com/simonswine/kube-latency

Benchmark kubernetes' Pod network
https://github.com/simonswine/kube-latency

grafana kubernetes latency network network-overlay prometheus

Last synced: about 1 month ago
JSON representation

Benchmark kubernetes' Pod network

• Host: GitHub
• URL: https://github.com/simonswine/kube-latency
• Owner: simonswine
• License: apache-2.0
• Created: 2017-03-26T17:49:34.000Z (over 7 years ago)
• Default Branch: master
• Last Pushed: 2017-04-25T14:16:09.000Z (over 7 years ago)
• Last Synced: 2024-11-02T07:14:08.610Z (about 1 month ago)
• Topics: grafana, kubernetes, latency, network, network-overlay, prometheus
• Language: Go
• Size: 10.5 MB
• Stars: 39
• Watchers: 3
• Forks: 10
• Open Issues: 1
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

• awesome-kubernetes - Kube Latency - This is a simple little tool, that helps to measure bandwidth and latency in a kubernetes Pod Network. (Tools)

README

        
# kube-latency

This is a simple little tool, that helps to measure bandwidth and latency in a

kubernetes Pod Network.

![Screenshot Grafana](contrib/screenshots/grafana.png)

## Tests

These tests are excuted on every Node (through a DaemonSet) every 10 seconds

* Test `/ping` all other Pods (5 times)

* Download `/data` (16MBytes) from one randomly selected node

The result of the tests is tagged with Zone, Node and Pod information and

scraped by Prometheus. The results can graphed using Grafana.

## Example test case

```

# Setup prometheues + kube-latency in cluster using helm

## prometheus runs on nodes that are labeled `role: monitoring`

## kube-latency runs on nodes that are labeled `role: worker`

helm upgrade --install --values contrib/test/kube-latency-values.yaml test-kl contrib/helm/kube-latency

# Setup grafana

helm upgrade --install --values contrib/grafana/helm-values.yaml grafana stable/grafana

# Port forwards

## prometheus localhost:9090

## grafana    localhost:9091

kubectl port-forward --namespace default $(kubectl get pods --namespace default -l "app=prometheus,release=test-kl" -o jsonpath="{.items[0].metadata.name}") 9090:9090

kubectl port-forward --namespace default $(kubectl get pods --namespace default -l "app=grafana-grafana,component=grafana" -o jsonpath="{.items[0].metadata.name}") 9091:3000

# add the datasource to grafana

# Import Grafana's dashboard from contrib/grafana/kube-latency-dashboard.json.

The username and password for Grafana is admin.

# Scale up the asg

for i in {6..100..3}; do

  echo "scaling to $i" && aws autoscaling set-desired-capacity --auto-scaling-group-name kubernetes-nonprod_devcluster-worker --desired-capacity $i && sleep 60

done

```