Problems installing with conda - HTTP 403 FORBIDDEN

[SamTygier]

Hi,

I am having trouble installing ipp with conda today. I see this on github actions, locally and in a docker container. For example, in a clean condaforge/miniforge3 docker image:

# conda create -v -n test2 -c intel ipp
Channels:
 - intel
 - conda-forge
Platform: linux-64
Collecting package metadata (repodata.json): ...working... info     libmamba Reading cache files '/tmp/tmpb6fsu7yj.json.*' for repo index 'installed'
info     libmamba Reading repodata.json file "/tmp/tmpb6fsu7yj.json" for repo installed
failed

UnavailableInvalidChannel: HTTP 403 FORBIDDEN for channel intel <https://conda.anaconda.org/intel>

The channel is not accessible or is invalid.

I get the same issue with mamba. Also the page https://anaconda.org/intel/ipp is no longer accessible.

Is this a temporary server issue, or has something moved to a new address?

Thanks,

Sam

[SamTygier]

Thank Jacob for finding a working repo.

Any word from Intel on what's happening?

[DevelVitoF]

How can I remove all pakages references to this Intel channel, like openvino package.

Can´t use my conda installation because off this unexpected removal off the Intel channel.

[DevelVitoF]

Ok fixed mysel

conda config --remove channels intel

conda remove --force intel-openmp

conda remove --force openvino stuff

[wens]

conda config --remove channels intel 

This command doesn't work on my server.

What I do is delete line '- intel' from file /opt/intel/oneapi/intelpython/.condarc ( this is vary depends on your installation)

add new channel

#conda config --add channels https://software.repos.intel.com/python/conda/

[AnurajKA]

Dear Team,

Is there any update on this issue ? 

Is this a temporary glitch ? 

So many workflows are failing due to this issue. 

Can someone kindly respond. ?

Certains versions which we are using in intel channel is not available in the latest repo(https://software.repos.intel.com/python/conda).
Eg: Intelpython 23.0.0 is not available 

Thanks 

Anuraj k A

[Louie_T_Intel]

Hi 

Please try to replace "-c intel" with " -c https://software.repos.intel.com/python/conda/" and you could find similar suggestion in AI Tools selector.

https://www.intel.com/content/www/us/en/developer/tools/oneapi/ai-tools-selector.html

We change it due to anaconda policy change. Sorry for inconvenience.

Hope it work for you.

[Alex-F]

Hi Louie,

To echo what Anuraj mentioned above, the new channel does not include all of the packages which the original intel channel hosted. It seems the older packages did not get migrated, which is preventing myself and other users from re-building environments.

As an example, here is a conda environment for an older project:

(my_env) $ conda list | grep intel
_libgcc_mutex             0.1                 conda_forge    intel
_openmp_mutex             4.5                       2_gnu    intel
common_cmplr_lib_rt       2021.3.0             intel_3350    intel
common_cmplr_lic_rt       2021.3.0             intel_3350    intel
dpcpp-cpp-rt              2021.3.0             intel_3350    intel
dpcpp_cpp_rt              2021.3.0             intel_3350    intel
icc_rt                    2021.3.0             intel_3350    intel
impi_rt                   2021.3.0              intel_294    intel
intel-cmplr-lib-rt        2021.3.0             intel_3350    intel
intel-cmplr-lic-rt        2021.3.0             intel_3350    intel
intel-opencl-rt           2021.3.0             intel_3350    intel
intel-openmp              2021.3.0             intel_3350    intel
intelpython               2024.1.0                      0    intel
opencl_rt                 2021.3.0             intel_3350    intel
tbb                       2021.12.0             intel_495    intel
zlib                      1.2.11                  intel_3    intel

 
This environment has various packages installed from the old intel conda channel. When searching the new channel using the conda CLI (as attempting to browse the channel online provides an XML error):

(base) $ conda search opencl_rt -c https://software.repos.intel.com/python/conda/ --override-channels
# (A long stack trace for a json paring exception, I think due to the channel metadata that gets cached)
# Name                       Version           Build  Channel
opencl_rt                   2022.2.0      intel_8734  python/conda
opencl_rt                   2022.2.1     intel_16953  python/conda
opencl_rt                   2023.0.0     intel_25370  python/conda
opencl_rt                   2023.1.0     intel_46305  python/conda
opencl_rt                   2023.2.0     intel_49495  python/conda
opencl_rt                   2023.2.3        intel_20  python/conda
opencl_rt                   2023.2.4     intel_49553  python/conda
opencl_rt                   2024.0.0     intel_49847  python/conda
opencl_rt                   2024.0.1     intel_49878  python/conda
opencl_rt                   2024.0.2     intel_49895  python/conda
opencl_rt                   2024.1.0       intel_963  python/conda
opencl_rt                   2024.1.2       intel_995  python/conda
opencl_rt                   2024.2.0       intel_981  python/conda

It appears any older packages are no-longer hosted.

Will older package versions which were available on the original intel channel (such as 2021.3.0) be hosted at a later date? To quote Jacob in this thread, "Removing the old packages breaks deployment of anything that requires them. If we can't trust this channel to keep the packages there, we can't use this compiler."

This issue appears to have resulted in a couple more forum posts, I'll link them below:

• https://community.intel.com/t5/oneAPI-Registration-Download/Using-conda-to-get-the-Intel-compiler-runtime-libraries/m-p/1618456 
• https://community.intel.com/t5/oneAPI-Registration-Download/Disappearance-of-intel-conda-channel/m-p/1614250 

[cleebp]

I am running into a very similar issue when trying to retrieve the older set of 2022.2.1 mkl packages via the intel conda channel, the workaround provided does not seem to resolve these older versions.

Echoing the need for the full archive of the older intel conda channel to be preserved, our automatic deployment and builds of packages like scipy are now broken because of this.

[Alex-F]

Hi cleebp,

In case you've not seen, Rahul (an Intel moderator) has added a comment here that seems to suggest that Intel is not planning to host older packages on their new channel.

This is very unfortunate news for myself. I am planning to look into alternative build methods for my codes outside of the conda eco-system as a result.

Related Tickets:

1. Disappearance of intel conda channel (Part 1)
2. Disappearance of intel conda channel PART 2
3. Re: Using conda to get the Intel compiler runtime libraries
4. "Old" Intel compilers and Conda

Alex

[Juhi_B_Intel · Accepted solution]

Hi Everyone!

Thank you for reporting that the Intel channel is no longer available at anaconda.com. For install Intel oneAPI libraries, compilers, and tools as stand-alone versions, please refer to this document: install-stand-alone-tools

Yes not all the older versions were posted in the new channel. Please note that Older versions do not have the latest functional and security updates and Users should update to the latest version of oneAPI Toolkits and components.
Despite that some older versions have been restored, if there is any particular one you need for your use please open a Priority Support Ticket with Intel. https://www.intel.com/content/www/us/en/developer/tools/oneapi/commercial-base.html#gs.f9mewg

[Olórin]

The champions at intel stroke again. Nice job. Fresh install, nothing works. It's not like I also have several issues with brand new one api install etc etc. Business as usuel like since fifteen years.

[pphili]

Hi,

I am wondering if there is a way to include install instructions from https://software.repos.intel.com/python/conda into a .yml file? 

I have tried a file like this

name: test

channels:

- defaults

- conda-forge

- https://software.repos.intel.com/python/conda/ # Intel channel for mkl

dependencies:

- python=3.10

- pip

- https://software.repos.intel.com/python/conda/::mkl-service # Explicitly from Intel channel

which leads to the following when I attempt to create the environemnt:

Collecting package metadata (repodata.json): failed

UnavailableInvalidChannel: HTTP 404 NOT FOUND for channel python/conda <https://conda.anaconda.org/python/conda>

The channel is not accessible or is invalid.

You will need to adjust your conda configuration to proceed.
Use `conda config --show channels` to view your configuration's current state,
and use `conda config --show-sources` to view config file locations.

Is there a different supported approach?