Kube-NetLag

Table of Contents

• Kube-NetLag
  • Overview
  • Features
  • Tested Kubernetes Distributions
  • Installation
    • Prerequisites
    • Deploy using Kubernetes Manifests
    • Deploy using Helm Chart
      • Install the Chart
  • Testing the Deployment
    • Verify that the DaemonSet is running
    • Check logs to confirm proper operation
  • Uninstall the Chart
  • Configuration
    • Global Parameters
    • Network & Ports
    • Prometheus Configuration
    • Resource Allocation
  • Prometheus Integration
  • Exposed Prometheus Metrics
    • Latency Metrics
    • Example Prometheus Query
  • Contributing
  • Code of Conduct
  • Disclaimer
  • License
  • Author

Overview

Kube-NetLag is a lightweight network performance testing and latency simulation tool deployed as a DaemonSet across your Kubernetes cluster. It helps measure network delays between nodes, making it useful for debugging and optimizing cluster communication.

You can deploy Kube-NetLag using either Kubernetes manifests or a Helm chart. The Helm chart simplifies deployment and configuration, enabling seamless monitoring with Prometheus integration.

Features

• ✅ Network performance testing using Netperf.
• 📊 Prometheus metrics exposure for monitoring.
• ⚙️ Customizable ports & environment variables for flexibility.
• 📡 Automatic Prometheus configuration (optional).

Tested Kubernetes Distributions

Kube-NetLag has been successfully tested on the following Kubernetes distributions:

• ✅ Rancher Kubernetes
• 🔄 More distributions to be tested...

If you tested it on another Kubernetes distribution, feel free to contribute and share your feedback!

Installation

Prerequisites

• Kubernetes 1.26+
• Helm 3.0+
• Prometheus (if monitoring is required)

Deploy using Kubernetes Manifests

The manifest files are located in the manifests directory.

kubectl apply -f manifests/

Deploy using Helm Chart

Install the Chart

To install Kube-NetLag with default values:

helm repo add alazidis https://your-helm-repo-url
helm install kube-netlag alazidis/kube-netlag

To install with a custom configuration:

helm install kube-netlag alazidis/kube-netlag -f values.yaml

Testing the Deployment

Verify that the DaemonSet is running:

kubectl get pods -n kube-netlag

Check logs to confirm proper operation:

kubectl logs -l app.kubernetes.io/name=kube-netlag -n kube-netlag

Uninstall the Chart

To completely remove Kube-NetLag, run:

helm uninstall kube-netlag
kubectl delete namespace kube-netlag

Configuration

The Helm chart allows full customization via the values.yaml file.

Global Parameters

Parameter	Description	Default
image.repository	Docker image repository	alazidis/kube-netlag
image.tag	Image tag	Chart.Version
image.pullPolicy	Image pull policy	Always
namespaceOverride	Overrides the namespace for deployment	""

---

Network & Ports

Parameter	Description	Default
ports.containerPort	Netperf service port	12865
ports.hostPort	Host-mapped Netperf port	12865
ports.containerPort	Prometheus metrics port	9090
ports.hostPort	Host-mapped Prometheus port	9090

Note: If modifying ports, update extraEnv accordingly.

---

Prometheus Configuration

Parameter	Description	Default
prometheusConfig.create	Enable automatic Prometheus ConfigMap	true
prometheusConfig.namespace	Namespace for Prometheus ConfigMap	""

---

Resource Allocation

Parameter	Description	Default
resources.requests.cpu	Requested CPU	200m
resources.requests.memory	Requested Memory	40Mi
resources.limits.cpu	CPU limit	400m
resources.limits.memory	Memory limit	80Mi

---

Prometheus Integration

If prometheusConfig.create is set to true, a Prometheus scrape job will be automatically created:

- job_name: "kube-netlag-daemon"
  kubernetes_sd_configs:
    - role: pod
      namespaces:
        names:
          - {{ .Values.namespaceOverride | default .Release.Namespace }}
  relabel_configs:
    - source_labels: [__meta_kubernetes_pod_label_app]
      action: keep
      regex: {{ include "..name" . }}
    - source_labels: [__meta_kubernetes_namespace]
      action: keep
      regex: {{ .Values.namespaceOverride | default .Release.Namespace }}
      .....

Prometheus will automatically collect network latency and performance metrics from Kube-NetLag.

Exposed Prometheus Metrics

Kube-NetLag provides the following Prometheus metrics to monitor network latency between Kubernetes nodes.

Latency Metrics

Metric Name	Description
node_min_latency_ms	Minimum latency in microseconds between nodes.
node_max_latency_ms	Maximum latency in microseconds between nodes.
node_avg_latency_ms	Average latency in microseconds between nodes.

Each metric includes the following labels:

• from_node – Name of the source node (The current Node).
• to_node – Name of the destination node.
• from_ip – IP address of the source node.
• to_ip – IP address of the destination node.

Example Prometheus Query

To visualize average latency between nodes in Prometheus:

node_avg_latency_ms{from_node="node-1", to_node="node-2"}

Contributing

We welcome contributions from the community! 🚀
If you'd like to report an issue, request a feature, or contribute code, please check out our:

• 📝 Contribution Guidelines
• 💬 Open an Issue
• 🔄 Submit a Pull Request

Code of Conduct

We follow the Contributor Covenant Code of Conduct.
Please be respectful and inclusive to all community members.

Disclaimer

Kube-NetLag is not an official Kubernetes project and is not affiliated with or endorsed by the Cloud Native Computing Foundation (CNCF) or Kubernetes.

License

Kube-NetLag is licensed under the Apache 2.0 License.
By contributing, you agree that your code will also be licensed under Apache 2.0.

🔗 Read the full license here: LICENSE

Author

• Apostolos Lazidis 🚀