Change of listening address to 0.0.0.0

[JakeJungJoonYoon]

What if we use 0.0.0.0 as the listening IP address of JupyterLab and VSCode? Those two processes run correctly on the container, but the IP address is configured as the loopback address, 127.0.0.1, and it's not receiving the requests. If it's not designed so in the first place, I can make a PR for you. Thanks.

[jkhenning]

Hi @JakeJungJoonYoon,

A PR is very welcome 🙏 - I think it will be easier to discuss once we have a fix suggestion 🙂

[JakeJungJoonYoon]

I made a PR #13 which can be accessed by below link, which is changing lookback address (127.0.0.1) to anywhere address (0.0.0.0).
#13

I've read your documentation carefully, and found out that you meant to use ssh tunneling. However, in our on premise K8S server, there is no service attached to the agent pod, so ssh tunneling is keep failing like below.

SSH tunneling failed, retrying in 3 seconds
Starting SSH tunnel


SSH tunneling failed, retrying in 3 seconds
Starting SSH tunnel


SSH tunneling failed, retrying in 3 seconds
Starting SSH tunnel


SSH tunneling failed, retrying in 3 seconds
Starting SSH tunnel

So, I suggest this for those who have same problem with me.

1. Clone github clearml-session repository.
2. Change those 127.0.0.1 address to 0.0.0.0, like I did in PR.
3. Attach service for the agent.
4. You can use NodePort or even, attach Ingress to service.

The K8S service yaml file is following.

kind: Service
apiVersion: v1
metadata:
  name: clearml-agent-ports
spec:
  ports:
  - name: clearml-agent-vscode
    port: 9000
    targetPort: 9000
    nodePort: 31100
  - name: clearml-agent-jupyterlab
    port: 8888
    targetPort: 8888
    nodePort: 31120
  - name: clearml-agent-ssh
    port: 10022
    targetPort: 10022
    nodePort: 31140
  type: NodePort    
  selector:
    app.kubernetes.io/instance: clearml-agent

Thanks.