克隆 conda env 失败：错误 conda.core.link:_execute(543)

Question

I'm trying to clone a conda environment in order to use it on a new machine. On first machine:

conda-env export -n dvina > dvina.yml

On target machine:

conda-env create -n dvina -f=dvina.yml

This fails with error conda.core.link:_execute(543) . Searching reveals numerous posts with conda.core.link , but none with 543 . This is the the bash shell output from the target machine:

$ conda-env create -n dvina -f=dvina.yml
Collecting package metadata: done
Solving environment: done

Downloading and Extracting Packages
numpy-base-1.15.4    | 4.2 MB    | ##################################### | 100% 
mkl_random-1.0.2     | 383 KB    | ##################################### | 100% 
mkl_fft-1.0.6        | 191 KB    | ##################################### | 100% 
gcc_linux-64-7.3.0   | 10 KB     | ##################################### | 100% 
pytz-2018.7          | 248 KB    | ##################################### | 100% 
gxx_linux-64-7.3.0   | 9 KB      | ##################################### | 100% 
graphite2-1.3.12     | 106 KB    | ##################################### | 100% 
ca-certificates-2018 | 124 KB    | ##################################### | 100% 
libcurl-7.62.0       | 517 KB    | ##################################### | 100% 
python-dateutil-2.7. | 274 KB    | ##################################### | 100% 
intel-openmp-2019.1  | 885 KB    | ##################################### | 100% 
gfortran_linux-64-7. | 9 KB      | ##################################### | 100% 
libgcc-7.2.0         | 304 KB    | ##################################### | 100% 
pandas-0.23.4        | 10.0 MB   | ##################################### | 100% 
numpy-1.15.4         | 47 KB     | ##################################### | 100% 
mkl-2019.1           | 204.6 MB  | ##################################### | 100% 
curl-7.62.0          | 143 KB    | ##################################### | 100% 
six-1.12.0           | 22 KB     | ##################################### | 100% 
openbabel-2.4.1      | 5.1 MB    | ##################################### | 100% 
binutils_linux-64-2. | 9 KB      | ##################################### | 100% 
libpng-1.6.35        | 335 KB    | ##################################### | 100% 
Preparing transaction: done
Verifying transaction: done
Executing transaction: failed
ERROR conda.core.link:_execute(543): An error occurred while installing package 'defaults::icu-58.2-h9c2bf20_1'.
IsADirectoryError(21, 'Is a directory')
Attempting to roll back.

Rolling back transaction: done

[Errno 21] Is a directory: '/home/mheller/anaconda3/pkgs/icu-58.2-h9c2bf20_1/lib/icu/current'

Both machines are running Ubuntu; the source machine 16.04.2, the target machine 16.04.5. I've updated conda on both: conda update -n base conda -c anaconda .

A folder called dvina is created, but only contains a folder conda-meta with json files.

As a note, this happens on two machines with Ubuntu 16.04.5, but works if I clone an environment on the source machine itself with Ubuntu 16.04.2. Also, I observe the same behavior if I try to create a new environment from scratch, but only on the Ubuntu 16.04.5 machines. My understanding is that I cannot roll back to Ubuntu 16.04.2, so how can I solve this issue from the conda side?

Answer 1

TL;DR:

use conda install icu -c conda-forge first.

Hey, friends, I have been troubled by this problem for months. Today it comes up again and I decided to solve it.

I think it may be a bug of icu side. I have seen the problem several times, and it all happens when installing icu = 58.2-h9c2bf20_1 .

Guess what? When I tried a new version of icu , it is solved!

I use the following command:

conda install icu -c conda-forge

The version of icu in conda-forge is about 60+ , and it can be installed without pain.

After installing icu , then I continue to install other packages, and it works!

Hope it works for you :)