Examples¶

When Singularity-CRI is installed and configured and kubelet is restarted, you may use Kubernetes as you usually do. Here we will show some examples so you can verify your installation and probably discover some interesting use cases.

All examples can be found in Singularity-CRI repo. Further assumed you are located in Singularity-CRI repository root.

Hello, Kubernetes!¶

This is a dummy HTTP server suitable to verify Kubernetes installation.

$ kubectl apply -f examples/k8s/hello-kubernetes.yaml

After pods are up you should be able to hit hello-kubernetes Kubernetes service and see a web page with “Hello, World!” greeting.

Cat images server¶

Here we will walk through a basic Kubernetes example with SIF. We will deploy http file server that listens on port 8080 and creates a Kubernetes service to make it public on port 80.

Note

To make Singularity-CRI pull image from cloud library an explicit cloud.sylabs.io prefix should be specified in image field.

To create a deployment and a service run the following:

$ kubectl apply -f examples/k8s/image-service.yaml

To verify objects are indeed created you can do:

$ kubectl get deploy && \
  kubectl get svc

If everything is fine you should be able to access the cats server through the Kubernetes service. that has just been created. There are couple URLs that where created for testing purposes, /cats/good is the one with cats.

Bookshelf service¶

This is service for storing and searching books with MongoDB as a storage. Consists of two parts: MongoDB and a back-end application that provides API.

$ kubectl apply -f mongo.yaml && \
  kubectl apply -f bookshelf.yaml

After that you should be able to interact with bookshelf Kubernetes service.

The API is the following:

List books

GET /books

Create new book

POST /books

{
"title": "Les Misérables",
"author": "Victor Hugo",
"published_date": "1862",
"description": "Examining the nature of law and grace, the novel elaborates upon the history of France, the architecture and urban design of Paris, politics, moral philosophy, antimonarchism, justice, religion, and the types and nature of romantic and familial love."
}

Update existing book

PUT /books/<id>

{
"title": "Les Misérables",
"author": "Victor Hugo",
"published_date": "June 1862",
"description": "Examining the nature of law and grace, the novel elaborates upon the history of France, the architecture and urban design of Paris, politics, moral philosophy, antimonarchism, justice, religion, and the types and nature of romantic and familial love."
}

Get existing book

GET /books/<id>

Delete existing book

POST /books/<id>:delete

Image recognition using NVIDIA GPU¶

Here we will deploy image recognition application that uses NVIDIA GPUs.

To create a deployment and a service run the following:

$ kubectl apply -f darkflow.yaml

To verify that objects are indeed created you can do:

$ kubectl get deploy && \
  kubectl get svc

If everything is fine you should be able to access Darkflow UI that is exposed with darkflow-web Kubernetes service.

Note

You may need to change serverURL value in a ConfigMao from the example above according to your cluster configuration. It should point to the darkflow-front Kubernetes service.

Also you can change input and output directories location on your host.