I am trying to build a dockerimage using the NVIDIA Merlin pytorch dockerfile available here.
I have logged in into nvcr.io using docker login nvcr.io and the creditentials generated from https://ngc.nvidia.com/setup/api-key, which are $oauthtoken and a custom generated key. I got the Login Succeeded message. When I try to build that dockerfile, I get this:
=> [internal] load build definition from Dockerfile 0.0s
=> => transferring dockerfile: 3.22kB 0.0s
=> [internal] load .dockerignore 0.0s
=> => transferring context: 2B 0.0s
=> resolve image config for docker.io/docker/dockerfile:1.2 1.2s
=> [auth] docker/dockerfile:pull token for registry-1.docker.io 0.0s
=> CACHED docker-image://docker.io/docker/dockerfile:1.2@sha256:e2a8561e419ab1ba6b2fe6cbdf49fd92b95912df1cf7d313c3e2230a333fdbcc 0.0s
=> [internal] load .dockerignore 0.0s
=> [internal] load build definition from Dockerfile 0.0s
=> CANCELED [internal] load metadata for nvcr.io/nvidia/tritonserver:23.06-py3 1.3s
=> CANCELED [internal] load metadata for nvcr.io/nvidia/pytorch:23.06-py3 1.3s
=> ERROR [internal] load metadata for nvcr.io/nvstaging/merlin/merlin-base:23.06 1.3s
=> [auth] nvstaging/merlin/merlin-base:pull,push token for nvcr.io 0.0s
------
> [internal] load metadata for nvcr.io/nvstaging/merlin/merlin-base:23.06:
------
Dockerfile:1
--------------------
1 | >>> # syntax=docker/dockerfile:1.2
2 | ARG MERLIN_VERSION=23.06
3 | ARG TRITON_VERSION=23.06
--------------------
ERROR: failed to solve: pull access denied, repository does not exist or may require authorization: server message: insufficient_scope: authorization failed
Is there something more I should do to obtain proper authentication? I've read the documentation here and all it says about logging in (in section 3 on pulling images as well as in section 1.4 about logging into docker) is "You must also have access and logged into the NGC container registry as explained in the NGC Getting Started Guide", linking to the "NGC Getting Started Guide", but the link is dead...
Or do I require nvidia-docker to access this repo? From the documentation I linked in the previous paragraph, it seems like docker is enough.