OpenI-Octopus: How to Avoid Resource Preemption in Kubernetes Clusters
· 9 min read
This article was firstly released at
Container Cubeon September 30th, 2020, refer to鹏城实验室启智章鱼教你彻底摆脱Kubernetes集群资源抢占难题
This article was firstly released at
Container Cubeon September 30th, 2020, refer to鹏城实验室启智章鱼教你彻底摆脱Kubernetes集群资源抢占难题
This article was firstly released at
Container Cubeon September 30th, 2020, refer to揭秘爱奇艺深度学习平台云原生迁移实践
The easiest way to deploy Volcano is using Helm charts.
Clone the repository to a local path:
# mkdir -p $GOPATH/src/volcano.sh/
# cd $GOPATH/src/volcano.sh/
# git clone https://github.com/volcano-sh/volcano.git
Official images are available on DockerHub. You can also build local images by running the following command:
cd $GOPATH/src/volcano.sh/volcano
make images
## Verify your images.
# docker images
REPOSITORY TAG IMAGE ID CREATED SIZE
volcanosh/vk-admission latest a83338506638 8 seconds ago 41.4MB
volcanosh/vk-scheduler latest faa3c2a25ac3 9 seconds ago 49.6MB
volcanosh/vk-controllers latest 7b11606ebfb8 10 seconds ago 44.2MB
NOTE: Ensure that the images are correctly loaded to your Kubernetes cluster. For example, if you are using kind luster, run the kind load docker-image <image-name>:<tag> command for each image.
| Organization | Contact (Github UserName) | Environment | Description |
|---|---|---|---|
| Baidu Inc | @tizhou86 | Production | The scheduler for offline training of the PaddlePaddle deep learning framework |
| Tusimple | @suleisl2000 | The scheduler for offline training of MXNet | |
| FfDL | @animeshsingh | ||
| MOGU Inc | @jiaxuanzhou | Production | The scheduler for offline training of tiny+ |
This document describes how to run kube-batch as a batch scheduler for Kubernetes. To get the complete code, go to master.