My understanding is that running build_image on a non-GCP machine where gcloud is available should work, but it fails due to this check hitting a google.internal url before launching the builder:
https://github.com/GoogleCloudPlatform/compute-gpu-installation/blob/main/linux/cuda_installer/__main__.py#L275-L281
My understanding is that running
build_imageon a non-GCP machine wheregcloudis available should work, but it fails due to this check hitting agoogle.internalurl before launching the builder:https://github.com/GoogleCloudPlatform/compute-gpu-installation/blob/main/linux/cuda_installer/__main__.py#L275-L281