No user in docker container when using `--build-template conda/pixi:v1`

[mashehu]

Not 100% sure if this is a wave problem, but since it didn't come up when using the default template, I think it is:

As in #971 I am working on converting nf-core modules to more automated seqera container builds.
Pipeline-wise everything runs correctly when using the default template. But if I switch to the pixi template, I get an error in the docker container:

File "/opt/wave/.pixi/envs/default/bin/drevalpy-make-hpam-yamls", line 10, in <module>
    >       sys.exit(hpam_split())
    >                ~~~~~~~~~~^^
    >     File "/opt/wave/.pixi/envs/default/lib/python3.13/site-packages/drevalpy/cli_run_cv.py", line 104, in hpam_split
    >       from drevalpy.models import MODEL_FACTORY, MULTI_DRUG_MODEL_FACTORY, SINGLE_DRUG_MODEL_FACTORY
    >     File "/opt/wave/.pixi/envs/default/lib/python3.13/site-packages/drevalpy/models/__init__.py", line 55, in <module>
    >       from .DrugGNN import DrugGNN
    >     File "/opt/wave/.pixi/envs/default/lib/python3.13/site-packages/drevalpy/models/DrugGNN/__init__.py", line 3, in <module>
    >       from .drug_gnn import DrugGNN
    >     File "/opt/wave/.pixi/envs/default/lib/python3.13/site-packages/drevalpy/models/DrugGNN/drug_gnn.py", line 8, in <module>
    >       import pytorch_lightning as pl
    >     File "/opt/wave/.pixi/envs/default/lib/python3.13/site-packages/pytorch_lightning/__init__.py", line 25, in <module>
    >       from lightning_fabric.utilities.seed import seed_everything  # noqa: E402
    >       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    >     File "/opt/wave/.pixi/envs/default/lib/python3.13/site-packages/lightning_fabric/__init__.py", line 35, in <module>
    >       from lightning_fabric.fabric import Fabric  # noqa: E402
    >       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    >     File "/opt/wave/.pixi/envs/default/lib/python3.13/site-packages/lightning_fabric/fabric.py", line 41, in <module>
    >       from lightning_fabric.loggers import Logger
    >     File "/opt/wave/.pixi/envs/default/lib/python3.13/site-packages/lightning_fabric/loggers/__init__.py", line 15, in <module>
    >       from lightning_fabric.loggers.tensorboard import TensorBoardLogger  # noqa: F401
    >       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    >     File "/opt/wave/.pixi/envs/default/lib/python3.13/site-packages/lightning_fabric/loggers/tensorboard.py", line 31, in <module>
    >       from lightning_fabric.wrappers import _unwrap_objects
    >     File "/opt/wave/.pixi/envs/default/lib/python3.13/site-packages/lightning_fabric/wrappers.py", line 33, in <module>
    >       from torch._dynamo import OptimizedModule
    >     File "/opt/wave/.pixi/envs/default/lib/python3.13/site-packages/torch/_dynamo/__init__.py", line 13, in <module>
    >       from . import (
    >       ...<6 lines>...
    >       )
    >     File "/opt/wave/.pixi/envs/default/lib/python3.13/site-packages/torch/_dynamo/aot_compile.py", line 16, in <module>
    >       from torch._dynamo.package import SystemInfo
    >     File "/opt/wave/.pixi/envs/default/lib/python3.13/site-packages/torch/_dynamo/package.py", line 1157, in <module>
    >       DynamoCache = DiskDynamoCache(os.path.join(cache_dir(), "dynamo"))
    >                                                  ~~~~~~~~~^^
    >     File "/opt/wave/.pixi/envs/default/lib/python3.13/site-packages/torch/_dynamo/package.py", line 1154, in cache_dir
    >       return cache_dir()
    >     File "/opt/wave/.pixi/envs/default/lib/python3.13/site-packages/torch/_inductor/runtime/cache_dir_utils.py", line 17, in cache_dir
    >       os.environ["TORCHINDUCTOR_CACHE_DIR"] = cache_dir = default_cache_dir()
    >                                                           ~~~~~~~~~~~~~~~~~^^
    >     File "/opt/wave/.pixi/envs/default/lib/python3.13/site-packages/torch/_inductor/runtime/cache_dir_utils.py", line 23, in default_cache_dir
    >       sanitized_username = re.sub(r'[\\/:*?"<>|]', "_", getpass.getuser())
    >                                                         ~~~~~~~~~~~~~~~^^
    >     File "/opt/wave/.pixi/envs/default/lib/python3.13/getpass.py", line 175, in getuser
    >       raise OSError('No username set in the environment') from e
    >   OSError: No username set in the environment

Here is the complete CI run: nf-core/drugresponseeval#80

Is this a wave bug?

[munishchouhan]

the problem is micromamba base image is pre-configured with a proper user environment, but the pixi image runs as root without setting the USER variable.
We will look into it

[mashehu]

Any updates on this issue, @munishchouhan?
it is currently blocking a wider adoption of our new pure seqera-container setup in nf-core.

[munishchouhan]

Any updates on this issue, @munishchouhan? it is currently blocking a wider adoption of our new pure seqera-container setup in nf-core.

@mashehu thanks for the reminder, is it possible to provide reproducer for this?

[mashehu]

is this good enough as a MRE:

The following script fails:

#!/usr/bin/env nextflow

nextflow.enable.dsl=2

process TEST_PYTHON_CONTAINER {
    container "python_pip_drevalpy:af9bf54ab78b9680"

    output:
    stdout

    script:
    """
    drevalpy-make-hpam-yamls \
       --model_name "NaiveDrugMeanPredictor"

    """
}

workflow {
    TEST_PYTHON_CONTAINER()
    TEST_PYTHON_CONTAINER.out.view()
}

and the config:

docker {
    enabled = true
    runOptions = '-u $(id -u):$(id -g)'
}

and the run command:

nextflow run test_container.nf -c docker-user.config

the container was created using:

wave --conda-file environment.yml --freeze --platform linux/amd64 -o yaml --build-template conda/pixi:v1

with the following content of environment.yml:

channels:
  - conda-forge
dependencies:
  - python=3.13
  - pip=25.3
  - pip:
      - drevalpy==1.4.1

It works if i don't use the docker.runOptions, but we need that in nf-core because of other things.

[munishchouhan]

fix: #974