NGC failsall calls with "Name or service not known"

Hello,

Running in Ubuntu 22.04 LTS on x86 processor with A1000 GPU.

I am trying to run a TAO is a self-hosted mode, but can never get past NGC commands. I have setup NGC in several forms – python virtual environment (so effectively native) and docker.

In each case, calls to NGC result in output like (added --debug):

os: Linux-6.8.0-40-generic
ngc-cli version: NGC CLI 3.53.0
[Errno -2] Name or service not known
Traceback (most recent call last):
  File "ngccli/ngc.py", line 163, in main
  File "ngcbpc/tracing.py", line 146, in wrapper_span
  File "ngcbpc/singleton.py", line 18, in __call__
  File "ngcbpc/tracing.py", line 83, in __init__
  File "ngcbpc/tracing.py", line 73, in is_enabled
  File "ngcbpc/util/ssh_utils.py", line 143, in is_port_open
socket.gaierror: [Errno -2] Name or service not known
Command: ngc registry model list nvidia/tao/gazenet:* --

I believe this is due to some network / DNS / proxy error. I am behind company proxy, and usually HTTPS_PROXY and similar environment variables are sufficient. That doesn’t seem to be the case here, and I’m otherwise not sure how to remedy this.

Is there any recommendation here? I haven’t found anything on the forums that directly suggests how to work around this. I’m trying to avoid any externally hosted cloud instances for the time being.

replying in to myself to (hopefully) keep the thread from going stale.

Is there any recommendation from nvidia/community or am I SoL here?

I have the same problem on Ubuntu 22.04.

Seems like I’m SoL, and that NVidia has no intention of resolving topics like this. Too bad, the TAO platform seemed interesting and useful but I can’t realistically evaluate for my company without resolving this issue. Seems from the forums that I’m not the only one with this problem, but no resolution on those threads either.