we-dcode / kubetunnel Goto Github PK

Clean autogenerated not in use files and change the project structure if needed

2. Add --proxy flags to kubetunnel-cli that sends their values through the CR spec.
   2.1. Add the flag itself - can be used multiple times [X]
   2.2. Parse the flag/s and send the values to the cr spec. [X]
   2.3 Add the DNS to the hosts file - send them to kubetunnel-server service. [X]

Currently it is a bit hard to understand the steps to reproduce in order to create a tunnel.

Especially because the commands are not specified to be for the local or for the remote environment

Having a step-by-step detailed documentation would help a lot.

1. Add sidecar for tcp proxy to kubetunnel deployment
   1.1 Need to choose proxy type to deploy
   1.2 Add proxy to helm chart of kubetunnel-server - only deploy it if needed with the relevant config from cr spec.

1. Query the application in controller and scaledown the original deployment, and update status in deployment with kubetunnel/lastReplicaCount.
2. Add logic for when disconnected/connected then scale back - scale down.
3. scale up - if replica count is not 0, then only remove label, else, scale up and remove label.

1. Add functionality from kubefwd package in cli
2. Option for specific namespaces or all
   svc name.namespace

Scenario:

Developer A works on Cart service
Developer B works on API Gateway service

The cart is a new functionality, never been deployed before. Developer A and Developer B want to integrate with each other.

1. Add command to deploy ad-hoc service with a tunnel
2. Add command to port-forward ONLY from Developer B to the cluster, so Developer B can access all existing services + the new adhoc service, but not tunnel anything

Add the new resolve all namespaces and proxy external private destinations features to the documentation.

Limitations:

1. Proxy doesn't support the same port multiple times, even on different dns
2. When using a proxy without graceful shutdown, the hosts file will have the real DNS records as proxy. Assuming the user can reach the DNS from his laptop as well (for example - Google PubSub or AWS SQS), the user will manually need to delete the host OR run the kubetunnel and gracefully shut down.
3. When using all namespaces flag, the service namespace will be the main DNS. In order to resolve all other namespaces the user will need to use a fully qualified URL.

1. Controller - add logic to query the application service account.
2. Add this to the values of the CR when a CR is created.
3. Check if need to add anything in kubetunnel-server helm chart to configure the serviceaccount.