intel / cloud-native-ai-pipeline

AI cloud native pipeline for confidential and sustainable computing
https://intel.github.io/cloud-native-ai-pipeline/
Apache License 2.0
37 stars 15 forks source link

scripts: easy tool to quick setup the cluster from zero #163

Closed leyao-daily closed 1 year ago

leyao-daily commented 1 year ago

This PR introduces an orchestrated procedure to setup a host machine and initialize or join a Kubernetes cluster. The script is designed to ensure host readiness, disable swap, configure kernel parameters, install necessary prerequisites like Docker and Kubernetes packages, setup Containerd, initialize or join the cluster, and optionally setup a Container Network Interface (CNI) plugin.

jiere commented 1 year ago

May I know why use Python to organize such scripts? Since we have "tools" directory also in this repo which hosts shell scripts there.

leyao-daily commented 1 year ago

May I know why use Python to organize such scripts? Since we have "tools" directory also in this repo which hosts shell scripts there.

Tools for the cnap related shell scripts, and this is used to setup k8s quick on a new setup machine, so i just move it to k8s-mainifests, one can execute it to setup or join cluster quick, especially for some testing or demo.

Using python can brings more feature and flexiblities to it. Maybe cross platform bulabula..

Of course, if we don't need this to setup k8s, i can move it to other repo or location. :)

jiere commented 1 year ago

Using python can brings more feature and flexiblities to it. Maybe cross platform bulabula..

Of course, if we don't need this to setup k8s, i can move it to other repo or location. :)

Yeah, what I am thinking is just about where to put such deployment related scripts, since our current demo repo is not really open sourced in public, but for real open source project, deployment related scripts are usually in Shell/Ansible/helm(yaml), etc.

leyao-daily commented 1 year ago

Using python can brings more feature and flexiblities to it. Maybe cross platform bulabula.. Of course, if we don't need this to setup k8s, i can move it to other repo or location. :)

Yeah, what I am thinking is just about where to put such deployment related scripts, since our current demo repo is not really open sourced in public, but for real open source project, deployment related scripts are usually in Shell/Ansible/helm(yaml), etc.

what do yout think a directoryutilities or dependencies, because k8s is indeed the dependency of the project?

leyao-daily commented 1 year ago

Current directory LGTM. Since we have involved KEDA/Kepler in this project scope in README.md, I will raise PR to include the Kepler deployment later on top of your work:) How about KEDA then? Maybe KEDA/Kepler scripts could be placed under tools/integration?

We have keda scable job for cnap yet, and for keda installation, we deliver the document under ./docs