Ray k8s cluster, cannot run new task when previous task failed

@architkulkarni , note that components is a private module of my project. In the workspace, I can actually import it. I do not need to install it. The following is the main directories of the code

# show workspace
$ pwd
/home/me/app
$ cd /home/me/app && tree
|-- docker
|-- myproject
|   |-- src
|   |   |-- components
|   |   |-- config
|   |   |   |-- algs
|   |   |   `-- envs
|   |   |-- controllers
|   |   |-- envs
|   |   |-- learners
|   |   |-- pretrained
|   |   |-- runners
|   |   |-- training_methods
|   |   |-- train_model_ray
|   |   `-- utils
|-- k8s
|   |-- depoly
|   |   `-- charts
|   |       `-- ray
|   |           |-- crds
|   |           `-- templates
|   `-- ray_cluster
|-- scripts
|   |-- cluster
|   |-- run_jobs
|   |-- submit_jobs
|   `-- tools
|   `-- utils

As you can see it, the components is in /home/me/app/myproject/components. Then in the workspace, I run the code

python myproject/src/main.py 

That is how can run the code.