Compiling C++ files, packaging in container and deployment

Ray Core
1

Hello,

Can you please help share pointers on how do users compile C++ code changes present in the workspace, and package code changes in the container for deployments in Kubernetes?

CC: @Dmitri

Thanks

2

@sangcho @ijrsvt
In other words, how should one compile local Ray core C++ changes and build the changes into a Docker image.

3

I think the way you can do this is to actually clone ray, build it locally Building Ray from Source — Ray v1.4.1, and package them as a container.

4

ok, thanks @sangcho , I will try those steps, can you please share sample Dockerfile that I can use to add C++ code changes to the container?

5

I’d probably just push the change to the github or other source control systems and checkout to that branch. Is what you are looking for re-building the wheel after the container is created?

6

I imagine you could build the wheel locally or wherever you’d like and have a Dockerfile that copies the wheel into the container and pip installs there.

7

Hello,I followed the documentation, I get the below error:

Exception information:
Traceback (most recent call last):
  File "/home/asmalvan/anaconda3/envs/ray/lib/python3.7/site-packages/pip/_internal/cli/base_command.py", line 180, in _main
    status = self.run(options, args)
  File "/home/asmalvan/anaconda3/envs/ray/lib/python3.7/site-packages/pip/_internal/cli/req_command.py", line 204, in wrapper
    return func(self, options, args)
  File "/home/asmalvan/anaconda3/envs/ray/lib/python3.7/site-packages/pip/_internal/commands/install.py", line 402, in run
    pycompile=options.compile,
  File "/home/asmalvan/anaconda3/envs/ray/lib/python3.7/site-packages/pip/_internal/req/__init__.py", line 85, in install_given_reqs
    pycompile=pycompile,
  File "/home/asmalvan/anaconda3/envs/ray/lib/python3.7/site-packages/pip/_internal/req/req_install.py", line 787, in install
    unpacked_source_directory=self.unpacked_source_directory,
  File "/home/asmalvan/anaconda3/envs/ray/lib/python3.7/site-packages/pip/_internal/operations/install/editable_legacy.py", line 46, in install_editable
    cwd=unpacked_source_directory,
  File "/home/asmalvan/anaconda3/envs/ray/lib/python3.7/site-packages/pip/_internal/utils/subprocess.py", line 244, in call_subprocess
    raise InstallationSubprocessError(proc.returncode, command_desc)
pip._internal.exceptions.InstallationSubprocessError: Command errored out with exit status 1: /home/asmalvan/anaconda3/envs/ray/bin/python -c 'import io, os, sys, setuptools, tokenize; sys.argv[0] = '"'"'/home/asmalvan/code/ray141/ray/python/setup.py'"'"'; __file__='"'"'/home/asmalvan/code/ray141/ray/python/setup.py'"'"';f = getattr(tokenize, '"'"'open'"'"', open)(__file__) if os.path.exists(__file__) else io.StringIO('"'"'from setuptools import setup; setup()'"'"');code = f.read().replace('"'"'\r\n'"'"', '"'"'\n'"'"');f.close();exec(compile(code, __file__, '"'"'exec'"'"'))' develop --no-deps Check the logs for full command output.
Removed build tracker: '/tmp/pip-req-tracker-u9n0vfp1'```

I am using RHEL8.4 to build
8

I have used ray base image to build entire ray as mentioned in documentation, I now have a build which is installed in the container:

Installed /ray/python
/home/ray/anaconda3/lib/python3.7/site-packages/setuptools/dist.py:692: UserWarning: Module Cython was already imported from /home/ray/anaconda3/lib/python3.7/site-packages/Cython/__init__.py, but /ray/python/.eggs/Cython-0.29.23-py3.7-linux-x86_64.egg is being added to sys.path
  pkg_resources.working_set.add(dist, replace=True)
/home/ray/anaconda3/lib/python3.7/site-packages/setuptools/dist.py:692: UserWarning: Module cython was already imported from /home/ray/anaconda3/lib/python3.7/site-packages/cython.py, but /ray/python/.eggs/Cython-0.29.23-py3.7-linux-x86_64.egg is being added to sys.path
  pkg_resources.working_set.add(dist, replace=True)

Successfully installed ray-2.0.0.dev0

Will try to commit this container and launch it on minikube cluster as next steps

9

Hello,

I managed to compile the entire ray source but still unable to see the changes deployed on the container.

  • I made few logger changes to file gcs_init_data.cc
  • Using instructions here I managed to compile Building Ray from Source — Ray v1.4.1 .cc file.
  • I made a wheel by running python setup.py bdist_wheel
  • When the created wheel is installed locally via pip install wheel-name.whl shows logger changes for gcs_init_data.cc in log file gcs_server.out
  • To deploy logger changes on the minkube, I built a new container and copied created wheel in the container.
  • The wheel is installed inside the container to be deployed by another pip install.
  • The new container when deployed onminikube does not show the logger changes that were made before building the container.

Can you please help, what I am missing here? happy to share more information

10

Before installing the wheel, make sure Ray is not installed in the container. That’s all I can think of at the moment.

1 Like