问题咨询

[yyds-xtt]

kubernetes.config.config_exception.ConfigException: Invalid kube-config file. No configuration found. 我在运行这个代码的时候碰到了这个错误，没有Kubernetes 的config配置文件。怎么解决呢？

[SidAg26]

Usually, the config file for K8s can be found at ~/.kube/config that is configured while setting up the cluster where the method config.load_kube_config by default looks for. Alternatives to this can be -

1. You may configure the environment variable KUBECONFIG with your file path e.g. export KUBECONFIG=/path/to/your/kubeconfig, OR,
2. Pass the file path to your config file explicitly e.g. config.load_kube_config('/path/to/your/kubeconfig')

[yyds-xtt]

How do we build our own Kubernetes cluster? Isn't this very troublesome? Are there any ready-made environments for us to use?

[yyds-xtt]

I hope to design more efficient reinforcement learning algorithms, however, I have spent a lot of time on environment configuration. Can you recommend some convenient environments?

[SidAg26]

The framework that you are accessing in this project requires OpenFaaS for serverless functions that execute over K8s. Thus, you need to -

1. Setup a K8s cluster, either a full deployment or any light-weight deployments such as microK8s
2. Install OpenFaaS on top of it
3. Ensure that OpenFaaS functions can be deployed over K8s - a good reference can be found here

However, if you only intend to develop Reinforcement Learning (RL) algorithms irrespective of the practical setup, you may not need to deploy the cluster and take the RL algorithms as a reference and work in any other Gymnasium compatible environment.