dvc import/get not working with a git repository with two remotes

Bug Report

Description

I have a git repository with two remotes, my .dvc/config is as follows, basically one remote for model, and one for data

[core]
    autostage = true
    remote = modelstore
['remote "modelstore"']
    url = s3://models/classifier/dvcstore
    endpointurl = <endpoint URL>
['remote "datastore"']
    url = s3://data/classifier/dvcstore
    endpointurl = <endpoint URL>

Here, model.dvc & data.dvc exist under /classifier dir

Then I have other repository to consume only the model from my artifact registry repo above. When I try to import the model I get the following error

dvc import <git repo> classifier/model --rev <tag>
ERROR: failed to import '<model path>' from '<git repo>'. - The path 'classifier/model' does not exist in the target repository '<git repo>' neither as a DVC output nor as a Git-tracked file.:

However the mode.dvc file exists in the repo under classifier/model. I have tried with and without a default remote, same error. Using dvc get, I only get ERROR: unexpected error

Reproduce

Expected

model artifacts are downloaded on local model

Environment information

Output of dvc doctor on consumer repo:

$ dvc doctor
DVC version: 2.27.2 (pip)
---------------------------------
Platform: Python 3.9.7 on Linux-4.18.0-372.26.1.el8_6.x86_64-x86_64-with-glibc2.28
Subprojects:
        dvc_data = 0.10.0
        dvc_objects = 0.4.0
        dvc_render = 0.0.11
        dvc_task = 0.1.2
        dvclive = 0.11.0
        scmrepo = 0.1.1
Supports:
        http (aiohttp = 3.8.3, aiohttp-retry = 2.8.3),
        https (aiohttp = 3.8.3, aiohttp-retry = 2.8.3),
        s3 (s3fs = 2022.10.0, boto3 = 1.24.59)
Cache types: reflink, hardlink, symlink
Cache directory: xfs on /dev/mapper/vg_li1744707733-lv_root
Caches: local
Remotes: None
Workspace directory: xfs on /dev/mapper/vg_li1744707733-lv_root
Repo: dvc, git

Output of dvc doctor on artifact registry repo:

$ dvc doctor
DVC version: 2.27.2 (pip)
---------------------------------
Platform: Python 3.9.7 on Linux-4.18.0-372.26.1.el8_6.x86_64-x86_64-with-glibc2.28
Subprojects:
	dvc_data = 0.10.0
	dvc_objects = 0.4.0
	dvc_render = 0.0.11
	dvc_task = 0.1.2
	dvclive = 0.11.0
	scmrepo = 0.1.1
Supports:
	http (aiohttp = 3.8.3, aiohttp-retry = 2.8.3),
	https (aiohttp = 3.8.3, aiohttp-retry = 2.8.3),
	s3 (s3fs = 2022.10.0, boto3 = 1.24.59)
Cache types: reflink, hardlink, symlink
Cache directory: xfs on /dev/mapper/vg_li1744707733-lv_root
Caches: local
Remotes: s3, s3
Workspace directory: xfs on /dev/mapper/vg_li1744707733-lv_root
Repo: dvc, git

[moisesrc13]

@pmrowla thanks so much!, adding the default remote in mode.dvc & data.dvc made it to work 🎉 if this is not yet in your documentation would be really useful to have

[pmrowla]

@moisesrc13 is it possible someone forgot to dvc push the latest data from your data registry?

Also, one thing you may want to consider is using the remote output option in your .dvc files, which enforces that the given file or dir should only be pushed to a specific remote.

So in classifier/model.dvc you would have:

outs:
  - md5: ...
    path: model
    ...
    remote: modelstore

and in classifier/data.dvc you would use remote: datastore

This way you can just run dvc push in your data registry (without manually specifying classifier/model or classifier/data and the relevant remote). This will also make dvc import use the correct remotes, without you needing to change the default remote in the data registry.

see: https://dvc.org/doc/user-guide/project-structure/dvc-files#output-entries