Isaac gym tutorial doesn't work

[1616ttkj]

“1.4.3. Inferencing with Pre-Trained Checkpoints” does not work with the following error.

2023-09-22 00:22:48 [10,364ms] [Error] [omniisaacgymenvs.utils.config_utils.path_utils] Invalid checkpoint path: omniverse://localhost/NVIDIA/Assets/Isaac/2022.2.1/Isaac/Samples/OmniIsaacGymEnvs/Checkpoints/cartpole.pth

“Nucleus check” looks fine.

[rthaker]

Hi @1616ttkj - The error message you’re seeing suggests that the path to the pre-trained checkpoint file is not valid or the file cannot be accessed. Here are a few things you can check:

1. Check the path: Make sure the path to the checkpoint file is correct. The path you provided seems to be correct according to the Isaac Sim documentation, but it’s always good to double-check. You can do this by navigating to the path in the Omniverse Nucleus browser and verifying that the cartpole.pth file exists.
2. Check Nucleus server connection: Ensure that your Isaac Sim is properly connected to the Nucleus server. You mentioned that the “Nucleus check” looks fine, but it might be worth checking the connection again just to be sure. You can do this by going to Window > Nucleus in Isaac Sim and checking the server status.
3. Check file permissions: It’s possible that Isaac Sim doesn’t have the necessary permissions to access the file. This is less likely if you’re using a local Nucleus server, but it’s still worth checking. You can check the file permissions in the Nucleus browser.
4. Try a different checkpoint file: If you have other pre-trained checkpoint files available, try using one of them to see if the problem is specific to the cartpole.pth file.

[1616ttkj]

I get the following result by Nucleus check.

http://omniverse-content-production.s3-us-west-2.amazonaws.com/Assets/Isaac/2022.2.1

I also changed localhost to this and it worked.
How can I achieve this on localhost?
Rewriting is a huge task.

[rthaker]

Hi @1616ttkj - Steps mentioned here will be useful to check nucleus through local host

Cannot find Nucleus User Management Options

I’m trying to setup a local nucleus server and add a couple users to it. I followed this tutorial and noticed that my interface is a bit different than what’s shown in the tutorial. Tutorial: Authentication and User Management — Omniverse Nucleus latest documentation Compared to the top 2 images of the tutorial, my omniverse login doesn’t have the “Create Account” button when logging in from a machine that’s different than the server. The nucleus navigator’s bottom left corner’s user manageme…

[1616ttkj]

Localhost is missing from NUCLEUS or NUCLEUS NAVIGATOR. Also, http://localhost:34080/ is empty in the browser.

[Simplychenable]

@1616ttkj just out of curiosity, was a localhost ever created? i believe the nucleus check should read omniverse://localhost if there is one set up (as opposed to the content-production.s3)

and from the sound of what you are describing, it may not have been there from the start. you can try setting it up via ‘Connect to a Server’ option under the Nucleus tab in your OV Launcher.

on the ‘Authentication’ screen, enter ‘localhost’. then, you should be able to enter your credentials that you had set up during OV installation. otherwise, you can try using ‘admin’ for both fields.

if you have any problems following, you can try watching the video on setting up localhost (starts at around 01:35) - Introduction — Omniverse Nucleus latest documentation

[1616ttkj]

This is my situation.
I could not create localhost.

[Simplychenable]

looks like that’s the root of the problem.

are you on a machine on a network or is this a personal machine? you could try uploading the OV Launcher log for the mods/devs to assess the issue; you can find the .log file here:

Windows - C:\Users\%username%\.nvidia-omniverse\logs\launcher.log
Linux - ~/.nvidia-omniverse/logs/launcher.log

[Simplychenable]

also, can you also make sure all services are running in the System Monitor? you can get there by navigating down to the Settings.

[1616ttkj]

The machine is Amazon AWS.
Service seems fine.

I think this log applies, but I’m not sure.

[2023-10-11 00:25:41.013] [debug] Failed to connect to the discovery service using port-based routing: (ws://localhost:3333/):  {
  target: {
    _events: {
      message: '[function] function i(e){t.call(this,new r(e,this))}',
      open: '[function] function l(){t.call(this,new s(this))}',
      error: '[function] function p(e){t.call(this,new o(e,this))}',
      close: '[function] function c(e,n){t.call(this,new a(e,n,this))}'
    },
    _eventsCount: 4,
    _binaryType: 'arraybuffer',
    _closeCode: 1006,
    _closeFrameReceived: false,
    _closeFrameSent: false,
    _closeMessage: '',
    _closeTimer: null,
    _extensions: {},
    _protocol: '',
    _readyState: 3,
    _receiver: null,
    _sender: null,
    _socket: null,
    _bufferedAmount: 0,
    _isServer: false,
    _redirects: 0,
    _url: 'ws://localhost:3333/',
    _req: null
  },
  type: 'error',
  message: 'connect ECONNREFUSED 127.0.0.1:3333',
  error: 'Error: connect ECONNREFUSED 127.0.0.1:3333\n' +
    '    at TCPConnectWrap.afterConnect [as oncomplete] (node:net:1494:16)'
}
[2023-10-11 00:25:41.015] [debug] Retrying to register an account in 5s, error:  Error: Failed to connect to the discovery service: localhost
    at /tmp/.mount_omniveosBZNd/resources/app.asar/dist/main.js:416:203284
    at process.processTicksAndRejections (node:internal/process/task_queues:95:5)
    at async /tmp/.mount_omniveosBZNd/resources/app.asar/dist/main.js:416:202749
[2023-10-11 00:25:46.015] [debug] Failed to connect to the discovery service using port-based routing: (ws://localhost:3333/):  {
  target: {
    _events: {
      message: '[function] function i(e){t.call(this,new r(e,this))}',
      open: '[function] function l(){t.call(this,new s(this))}',
      error: '[function] function p(e){t.call(this,new o(e,this))}',
      close: '[function] function c(e,n){t.call(this,new a(e,n,this))}'
    },
    _eventsCount: 4,
    _binaryType: 'arraybuffer',
    _closeCode: 1006,
    _closeFrameReceived: false,
    _closeFrameSent: false,
    _closeMessage: '',
    _closeTimer: null,
    _extensions: {},
    _protocol: '',
    _readyState: 3,
    _receiver: null,
    _sender: null,
    _socket: null,
    _bufferedAmount: 0,
    _isServer: false,
    _redirects: 0,
    _url: 'ws://localhost:3333/',
    _req: null
  },
  type: 'error',
  message: 'connect ECONNREFUSED 127.0.0.1:3333',
  error: 'Error: connect ECONNREFUSED 127.0.0.1:3333\n' +
    '    at TCPConnectWrap.afterConnect [as oncomplete] (node:net:1494:16)'

[Simplychenable]

ideally, we should see all services green and running. i believe the fact that you couldn’t authenticate a localhost before is because the AUTH service has errored out and stopped. if you can upload the launcher log in its entirety (as opposed to the snippet) to this thread, it would be helpful.

and unfortunately i do not have much experience working with a VM, so i would suggest following the mods/devs direction. generally speaking, ports could be a factor. while you wait for their feedback, you can check them in the doc here (note the warning and notes throughout the page) and also verify the outbound connectivity requirements at the bottom of the page.:

https://docs.omniverse.nvidia.com/nucleus/latest/ports_connectivity.html

[1616ttkj]

I was able to successfully create localhost after restarting all services.
Thank you very much.

[Simplychenable]

やった! hopefully the tutorials work for you now 🤞