Installing Anaconda on Linux Mint 20

If you started studying programming or became interested in it in recent years, you have probably heard of data science or machine learning. And you may also have heard of, or worked with, the R language, for its great power in statistical analysis, or Python, for its great versatility. Today’s article is about Anaconda, a distribution that aims to make it easier to install and manage tools for these two languages, especially for scientific applications.

If you prefer video, click the player below (in Portuguese). The full article follows the video.

Getting the installer

#

The Anaconda project website is simple and intuitive. Here we will cover installing the Individual Edition of Anaconda, which can be downloaded here. In this article, we will install the Python 3 installer for Linux systems. At the time of writing, the name of the downloaded installer file is Anaconda3-2020.07-Linux-x86_64.sh.

Installing the dependencies required for the GUI

#

Our goal here is to install Anaconda on a Linux distribution, more specifically Linux Mint 20. According to the instructions on the website, some dependencies are needed for the applications to run in graphical mode. For Debian-based distributions, as in our case, we need to install:

sudo apt install libgl1-mesa-glx libegl1-mesa libxrandr2 libxss1 libxcursor1 libxcomposite1 libasound2 libxi6 libxtst6

Verifying the integrity of the installer

#

It is always a good idea to check that the download of an installer completed successfully, to avoid starting the installation with corrupted files that can break system dependencies. If you have never heard of checksums and cryptographic hashes, I recommend this reading.

The project website has a list of the hashes of all of its installer files. At the time of writing, the hash of the Python 3 installer for Linux is 38ce717758b95b3bd0b1797cc6ccfb76f29a90c25bdfa50ee45f11e583edfdbf, but you should check the website, since a new version may be available by the time you read this article.

To check whether the downloaded file is corrupted, we can either compare the hash of the downloaded file with the expected hash by eye, or have the system do the comparison. For the visual check, we run the sha256sum command followed by the file name and compare the result with the hash on the website:

$ sha256sum Anaconda3-2020.07-Linux-x86_64.sh

To have the system compare the two hashes:

$ echo "38ce717758b95b3bd0b1797cc6ccfb76f29a90c25bdfa50ee45f11e583edfdbf Anaconda3-2020.07-Linux-x86_64.sh" | sha256sum --check

If the output of the command is OK, the hash of the file matches the given hash.

Creating an installation folder accessible to all users

#

By default, the Anaconda installer uses the user’s home folder (~/user-name). There is nothing wrong with that. However, if the machine is used by more than one user, it is sometimes better to have a single installation that serves all of them than to install it in the home folder of each user. So, before starting the installation, we will create a folder accessible to all users of the system, so that we can point the installer to it. Even if your machine is currently used only by you, I recommend keeping the installation in a system folder so that it can accommodate multiple users in the future.

On Linux, there is a standard hierarchy of system files (FHS, for Filesystem Hierarchy Standard) that recommends which folders to use in each situation. According to this standard, the opt folder is meant for programs that are not part of the operating system. So, we will create a folder for the installation inside opt. It is also important that this folder has the correct permissions so that the installer can write to it. So, creating an anaconda folder in opt:

sudo mkdir /opt/anaconda && sudo chmod ugo+w /opt/anaconda

Installation

#

Now we can perform the installation itself. Assuming you are in the folder where the installer was downloaded:

bash Anaconda3-2020.07-Linux-x86_64.sh -u

The -u option is requested by the installer when the installation is done in an existing folder. Just follow the on-screen instructions, accept the license terms, and change the installation folder to the one we created.

It is also important that, at the end of the installation, you confirm the option for the installer to initialize the conda manager. This way, the installer will update the .bashrc file so that you can start the conda manager directly from the terminal.

Verifying that the installation worked

#

After the installation finishes, close and reopen your terminal. If the installation succeeded, the (base) indicator will appear before the prompt:

(base) $

This indicates that the conda base environment was activated at startup (the default behavior). You can check which Python interpreter is being used with:

(base) $ which python
/opt/anaconda/bin/python

To deactivate the environment:

(base) $ conda deactivate
$

Now, the active Python interpreter is the system’s:

$ which python
/usr/bin/python

Making the installation accessible to all users

#

Earlier, we created a system-level folder so that all users could access it. Now we need to let other users actually use the Anaconda installation in that folder. First, let’s create a user group called anaconda:

$ sudo groupadd anaconda

Now, let’s transfer ownership of the installation folder to this group:

$ sudo chgrp -R anaconda /opt/anaconda

Let’s allow only the owner, root, and the group to have read and write permission:

$ sudo chmod 770 -R /opt/anaconda

Now, add whichever users you want to this anaconda group so that they can use the installation:

$ sudo adduser user_name anaconda

In case you don’t remember, at the end of the installation we allowed the installer to initialize conda. So, the installer made changes to the .bashrc file of the user who ran the installation so that conda is initialized every time the terminal is opened. We need to apply these changes to all users. The simplest way is to put the instructions in the /etc/profile file:

# >>> conda initialize >>>
# !! Contents within this block are managed by 'conda init' !!
__conda_setup="$('/opt/anaconda/bin/conda' 'shell.bash' 'hook' 2> /dev/null)"
if [ $? -eq 0 ]; then
    eval "$__conda_setup"
else
    if [ -f "/opt/anaconda/etc/profile.d/conda.sh" ]; then
        . "/opt/anaconda/etc/profile.d/conda.sh"
    else
        export PATH="/opt/anaconda/bin:$PATH"
    fi
fi
unset __conda_setup
# <<< conda initialize <<<

NOTE: It is important that you copy exactly what was added to your .bashrc file. Here I simply copied what was added to mine; you should copy what was added to yours.

The first time a new user opens the shell, they will need to run the conda init command and close it. That’s it: they will be able to use the installation normally the next times they open the shell, without needing that command.

Conclusion

#

The Anaconda installation itself is simple, but some details need to be covered for a multi-user installation. We covered those details in this article to help those who need it. We also have a more detailed article on virtual environments where we write about conda and its features. Don’t miss that article.

If you have questions, leave a comment. Share the article with your friends and see you in the next article.