You can mirror any channel from Anaconda Cloud using Anaconda Enterprise Repository.
In addition to the default Anaconda channel, some common channels are the Conda-Forge and R channels.
You can mirror these or any other channel into their own channels on your Anaconda Enterprise Repository installation.
1. Create a yaml file for the channel to be mirrored, r.yaml and conda-forge.yaml.
Add the channels and URL to the channel in each file. This configuration will mirror everything from the channel including all platforms and all python versions.
conda-forge.yaml:
channels:
- https://conda.anaconda.org/conda-forge
r.yaml:
channels:
- https://conda.anaconda.org/r
To start the mirroring process, run the following commands:
anaconda-server-sync-conda --mirror-config r.yaml --account=r-channel
anaconda-server-sync-conda --mirror-config conda-forge.yaml --account=conda-forge
--mirror-config : points to the path of the yaml file. If this file is not in the current working directory you will need to specify the full path.
--account: create the channel on your installation of Anaconda Repository. The name is specified after the = symbol.
2. Edit the yaml file to configure your mirror for specific packages, python versions, operating system platforms, architectures, and licenses.
platforms : specify what operating systems and architectures you want to mirror packages for: win-32, win-64, linux-32, linux-64, osx-32, osx-64.
python_versions: specify the python version of the packages being mirrored in the format of X.Y: 2.6, 2.7, 3.4, 3.5, etc.
pkg_list: specify the exact package names you want to mirror from the channel. Packages named here will be mirrored.
Example yaml file:
This configuration mirrors Linux 64-bit and Windows 64-bit platform versions of the Numpy and Scipy packages for Python-2.7 from the Conda-Forge channels:
channels:
- https://conda.anaconda.org/conda-forge
platforms:
- win-64
- linux-64
python_versions:
- 2.7
pkg_list:
- numpy
- scipy
MORE INFO: For other options like blacklisting, whitelisting, and license filtering, see the documentation.