Skip to content

[BUG] Docker example problems on ARM64 and Mac Silicon #5210

Description

@c-barakat

Tested with version 2.8.0
Trying to run the docker example on arm64 and Mac silicon systems as described in the README under examples/docker fails at multiple steps. I've documented here some of the changes that I made to the installation to be able to run the example locally, but it still fails when i try to run it on two separate machines.

Steps to reproduce the installation problems:

  1. Clone the repository
  2. git checkout tags/2.8.0
  3. Install NVFlare pip install -e .[dev,PT]
  4. You receive error ERROR: Could not find a version that satisfies the requirement tenseal==0.3.15; python_version < "3.13" and extra == "dev" (from nvflare-nightly[dev,pt]) (from versions: 0.3.17)
  5. This also occurs when building the docker job image since it also tries to pip install -e .[dev,MONITORING,PT]

Expected behavior:
The example should work out of the box or the README should reflect the necessary changes depending on the architecture.

Desktop:

  • OS: Ubuntu 24.04.4 LTS
  • Python Version 3.12.13
  • NVFlare Version 2.8.0

Additional context:
Above are only issues related to setting up the example.
However, when running the PyTorch example, further issues arise:

  • After running nvflare job submit jobs/hello-pt-docker --startup-kit path/to/startup/kit you will get ERRNO13: Permission denied as torchvision tries to download and extract the cifar10 dataset.
  • This is resolved by adding the following to both dockerfiles:
    RUN useradd --badname 1000
    RUN mkdir -p /var/tmp/nvflare && chmod -R 1777 /var/tmp/nvflare 
    
  • With these changes, this example will run locally but when trying to run between two machines another Permission denied issue appears RuntimeError: cannot connect to Docker daemon: Error while fetching server API version: ('Connection aborted.', PermissionError(13, 'Permission denied'))
  • As of this moment, I still cannot locate this issue and would appreciate any pointers as to how to resolve it or where to begin.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

Labels

bugSomething isn't working

Type

No type

Projects

No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions