https://github.com/cdayz/k8s-image-pre-puller

K8S Operator that ensures that docker images will be pre-pulled on all cluster nodes
https://github.com/cdayz/k8s-image-pre-puller

docker-image k8s k8s-operator pre-pull

Last synced: 3 months ago
JSON representation

K8S Operator that ensures that docker images will be pre-pulled on all cluster nodes

• Host: GitHub
• URL: https://github.com/cdayz/k8s-image-pre-puller
• Owner: Cdayz
• License: apache-2.0
• Created: 2024-02-23T18:14:11.000Z (11 months ago)
• Default Branch: master
• Last Pushed: 2024-05-20T22:35:58.000Z (8 months ago)
• Last Synced: 2024-10-11T06:42:24.705Z (3 months ago)
• Topics: docker-image, k8s, k8s-operator, pre-pull
• Language: Go
• Homepage:
• Size: 211 KB
• Stars: 3
• Watchers: 1
• Forks: 0
• Open Issues: 6
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# K8S Image Pre Puller

![CI](https://github.com/Cdayz/k8s-image-pre-puller/actions/workflows/main.yml/badge.svg)

It's a simple kubernetes operator that will help you automatically pre-pull docker images on nodes.

This operator can be used for speed-up pod initialization in cluster.

## How to use

### Install

You can install this operator by installing it through [Helm Chart](./charts/k8s-image-pre-puller/)

### Create PrePullImage

When operator installed in cluster you can simply create resources like this:

```yaml

apiVersion: images.cdayz.k8s.extensions/v1

kind: PrePullImage

metadata:

  name: my-cool-image-pre-pull

  namespace: my-namespace

spec:

  image: "gcr.io/my-cool-image"

```

Or you can specify `nodeSelector` to pre-pull image only on particular nodes.

```yaml

apiVersion: images.cdayz.k8s.extensions/v1

kind: PrePullImage

metadata:

  name: my-cool-image-pre-pull

  namespace: my-namespace

spec:

  image: "gcr.io/my-cool-image"

  nodeSelector:

    ssd: "true"

    feature.nvidia-runtime: "true"

    ...

```

This operator automatically will create a `DaemonSet` for create pre-pulling pods on every existing node.

**IMPORTANT**: This operator creates a `DaemonSet` and it will create `Pod`'s in cluster - this pods will consume resources according to configuration. It's important to have monitoring about this resources.