Troubleshooting Instructions

TROUBLESHOOTING LESSONS (Copy of LP)

Table of Contents:

1. Kernel Error / Jupyter Missing "Python (dojo-env)" in kernels
2. conda activate error - GitBash
3. "code" Command Not Working (Mac)
4. GitBash Error: Could Not Fork Child Process
5. Installing Graphviz (Windows)
6. Reinstalling Dojo-Env
7. Uninstalling Anaconda (Completely)
8. GitBash Opens in the Wrong Folder

---

1. Kernel Error / Jupyter Missing "Python (dojo-env)" in kernels

PROBLEM:

• If you see a message like this when trying to open the environment testing notebook:
  

• OR if "Python (dojo-env)" doesn't show up in Jupyter Notebook in EITHER:
  • The "New" menu on the Files tab in Jupyter.
• 
  

  • or the Kernel > Change Kernel menu inside a Jupyter Notebook:
    

    

Solution

• You missed a command after running the conda env create command during step 2, so let's rerun the command.
• First, shut down Jupyter Notebook
  • The quickest way is to go back to the Terminal/GitBash window that you used to start the notebook and press "control + C"

• In your terminal/GitBash, make sure it says (dojo-env) above or next to your terminal prompt.
  • if not, on Mac run:
    • "conda activate dojo-env"
  • on Windows run:
    • "source activate dojo-env"
• Then run the command below to install dojo-env as a kernel in jupyter:

python -m ipykernel install --user --name dojo-env --display-name "Python (dojo-env)"

Then, go ahead and boot up jupyter notebook and try again. It should now know the "Python (dojo-env)" kernel.

---

2. conda activate error - GitBash

PROBLEM:

• You see the following error message in GitBash:

CommandNotFoundError: Your shell has not been properly configured to use 'conda activate'.
To initialize your shell, run
    $ conda init <shell_name>
Currently supported shells are:
  - bash
  - fish
  - tcsh
  - xonsh
  - zsh
  - powershell
See 'conda init --help' for more information and options.</shell_name>

Solution 1:

• For your computer, run "source activate dojo-env" instead of conda activate.
  • ANYTIME you see a set of instructions that says to run "conda activate" you should ALWAYS replace the word "conda" with "source".

source activate dojo-env

PROBLEM - V2:

• If you are seeing this message EVERY time you open a new GitBash Window.

Solution 2:

• During step 2, you were told to run the following echo conda activate command but your computer requires you to use "source activate" instead of "conda activate"

## INCORRECT What you ran previously
echo "conda activate dojo-env" >> ~/.bash_profile

• Step 1) To fix this, you need to open the hidden file ".bash_profile" in VS Code to fix the command.
  • In GitBash run the following command to open the file in VS Code.

code ~/.bash_profile

• If this doesn't work and says file not found, change the "~./" to your full file path to your user folder:
  • In the example below, replace <USERFOLDER> with your actual User account name.

    
    

    • Note: if you do not know your username, run the "whoami" command in GitBash. The text displayed is your <USERFOLDER>

code /c/Users/<USERFOLDER>/.bash_profile

• Step 2) in the vs code window that opens, you should see the 3 lines of text below.
  

conda activate dojo-env
alias jnb="jupyter notebook"
alias lab="jupyter lab"

• Replace the word "conda" with "source" and then click on File > Save.
• So your final .bash_profile should now say:

source activate dojo-env
alias jnb="jupyter notebook"
alias lab="jupyter lab"

(Note: it is okay if there is additional text in this file. Whats important is that you change the conda activate command to source activate.)

• Step 3) Open a NEW GitBash window and the message should no longer appear!

---

3. "code" Command Not Working (Mac)

• If you are trying to open VS Code using the "code" command in your Terminal and receive an error message that says

command not found: code

• Open VS Code manually (check the Applications folder or use Spotlight Search)
• Open the Command Palette.
  • On the menu bar (top of the screen), select "View" > and then select "Command Palette"

• A popup window will appear with a ">" prompt.

• Start typing "install code" and you should see the option appear for "Shell Command: Install 'code' command in PATH".
  • Click on this option.

• It will then notify you that you will be prompted to give "osascript" administrative privileges.

• Click OK and then on the next pop-up window, enter your normal password for your Mac.
  • You should see a success message like the one below.

• You are all set! Try the code command again.

• NOTE: if you get an error message when running the install code command:
  • First UNinstall the code command (which is counter-intuitive).
  • Then rerun the install code command.

---

4. GitBash Error: Could Not Fork Child Process

The Problem:

Opening a GitBash windows displays an error message "Could not fork child process"

The Solution

Solution 1: (adapted from: https://rotadev.com/git-bash-error-could-not-fork-child-process-there-are-no-available-terminals-1-dev/ )

• The problem occurs on Windows with Git Bash when the Git Bash console is closed/killed without using ‘exit’.
  • 1. Close all Git Bash windows.
  • 2. Open Task manager (Keyboard shortcut: Control + Shift + Escape.)
  • 3. Find the ‘Git for Windows’ process.
    • If you do not see any "Git for Windows" processes, look for:
      • python
      • msys2
  • 4. Kill any processes that are running that meet the above criterion.
  • 5. Re-open Git Bash. The problem should be fixed.
  • See the Prevention heading below for how to avoid this error in the future

Solution 2 (if #1 did not resolve the issue):

• The problem may also occur due to VS Code's integrated terminal.
• Close any Visual Studio Code windows
• Check Task Manager for any Visual Studio Code processes that are still running and kill them.
• Re-open Git Bash. The problem should be fixed.
• See the Prevention heading below for how to avoid this error in the future

Solution 3 (will always work but not convenient)

• If neither solution worked, restart your computer.
• While not convenient, it will solve the issue. See the Prevention heading below for how to avoid this error in the future.

Prevention

• Always make sure to shut down jupyter notebook properly before closing.
  • From jupyter's file view:
    • Click Shutdown on the top right corner
  • From GitBash:
    • Press "Control + C" to quit jupyter server from the GitBash window that started jupyter notebook.
      • If asked y/n, answer "y".

---

5. Installing Graphviz (Windows)

Graphivz is an optional package that will allow you to generate a more advanced version of scikit-learn decision tree plots.

• Please follow these instructions from Graphivz's Website: https://forum.graphviz.org/t/new-simplified-installation-procedure-on-windows/224
• You will be downloading the installer from https://graphviz.org/download/
• Make sure to follow step "10 Select Add Graphviz to the system PATH for current user in this dialog:"

---

6. Reinstalling Dojo-Env

What to do if your environment breaks and you need to re-install it.

• It is not uncommon to accidentally break our virtual environment by installing a new package or updating a pre-existing one.
• In the event that your environment stops working and it needs to be re-installed:
  1. open your terminal/gitbash and deactivate your dojo-env:
     • Type conda activate base or conda deactivate and press enter.
     • Your terminal should now say (base) with your prompt instead of (dojo-env).
  2. Remove the broken dojo-env using the command:
     • conda remove --name dojo-env --all
     • enter y to approve the removal of the environment and hit enter.
  3. Wait for the env to be removed.
     • This will delete all of the files associated with JUST our dojo-env. So anconda will still be installed, we will just need to re-install our dojo-env.
  4. Once its completed, repeat the environment installation commands from step "2.3 Create the dojo-env environment"
     1. Don't forget to run the python -m ipykernel command after creating the dojo-env in step 2.3!

---

7. Uninstalling Anaconda (Completely)

When facing stubborn environment problems, sometimes the best course of action is to completely uninstall the environment AND Anaconda and to start over.

Pre-Uninstallation Verification Step:

• If you share your computer with another User who also uses Python:
  • Pause here and check with them BEFORE you uninstall anaconda. You will be removing all of their python environments too, even though they have a separate User account.

• If you share your computer with someone and they have concerns about uninstalling anaconda:
  • Stop here (for now).
    • Do not move forward with the instructions until you have spoken with your instructor.

Official Steps for Fully Uninstalling Anaconda:

The following steps are taken directly from the Official Uninstalling Anaconda documentation page, specifically "Option B. Full uninstall using Anaconda-Clean and simple remove."

• Install the Anaconda-Clean package from Anaconda Prompt (terminal on Linux or macOS):

conda install anaconda-clean

• In the same window, run the following command:

anaconda-clean --yes

• Once the process has been completed, manually delete any "anaconda3" or "anaconda2" folders that still exist.
  • It may be located in one of several possible folders. Run the following "ls -a" commands until you see a folder called "anconda2" or "anaconda3".
  • Once you see an anaconda folder, take note of:
    • Which command showed the folder.
      • Specifically, what did the command say after "ls -a"
      • We will refer to this as your "base folder" in the final step.
    • If the anaconda folder was "anconda2" or "anaconda3"
      • We will refer to this as your "anaconda folder" in the final step.
  • and jump to the very last command at the bottom of the page, with those 2 pieces of information.

ls -a ~/
ls -a ~/opt/
ls -a /opt/

• Run the final command to remove the anaconda folder once you've identified your "base folder" and "anaconda folder".
  • Replace {base_folder} with the actual folder name
  • Replace {anaconda_folder} with the actual folder name.

rm -rf {base_folder}{anaconda_folder}

Once you've replaced the placeholder folder names with your actual folder names, the command should look something like this:

rm -rf ~/opt/anaconda3
# or 
rm -rf ~/anaconda2
# or 
rm -rf ~/opt/anaconda3

Final Verification Step:

• Now, open a new terminal window and try running the "conda" command again. Your terminal should say that conda is not found.
  • If it says conda is not found, you are now ready to jump back up to the "Step 3 Commands" header above.
  • If your computer still displays a list of conda commands: please see the final portion of the Official Uninstallation Instructions "Removing Anaconda from .bash_profile", which is illustrated below:

• Open the two settings file for your terminal and remove anything related to anaconda:
  • 1) Run the "open ~/.bash_profile" (without quotation marks) command and a text editor window should open.
    • Examine the contents of this file and delete any lines that look like either of the screenshots below:
      • 
        
      • 
    • Save the file and close it.
  • 2) Now repeat the process, but use the "open ~/.zshrc" (without quotation marks) command
    • Delete any lines of code that look like the screenshots above.
    • Save the file and close it.

• FInal Verification:
  • Open a New Terminal Window (the changes you made above only take effect when opening a new window)
  • Run the "conda" command again and you should now see a message that says "conda not found"
    • If you are still seeing the list of conda commands instead, re-open the two files listed above and make sure you saved them after deleting the lines of code.
    • Close your terminal window and open a new one and try the "conda" command again.

• If the conda command still displays the list of commands after the steps above:
  • Try restarting your computer and attempting to run the command one more time.
  • If you are still seeing conda commands:
    • You should post your problem on the ds-python-installation discord channel for a TA or instructor to assist you.
    • You can also email your instructor about the issue (please cc: [email protected])

---

8. GitBash Opens in the Wrong Folder

PROBLEM:

Opening GitBash from Windows start menu (not from Github Desktop) and starting jupyter notebook displays a folder containing bin/cmd/dev/etc ( see screenshot below) instead of the expected User folder with Documents/Downloads/etc.

Cause

Your GitBash program defaults to the folder where GItBash is installed. What you are seeing above is the contents of C:/Program Files/GitBash. This happens to a small percentage of windows users.

Solution/Workaround

• First, run the following echo command to add one more alias to GitBash and then open a new GitBash window

echo 'alias ~="$HOME"' >> ~/.bash_profile

• Now test the command by runing the commands below to change directory to "~" and to display the current folder name.

cd ~
ls

• If you can see your Downloads, Desktop, Documents, etc. folders then the alias worked!

New Workflow

• From now on, if you want to open a GitBash in your user folder, you will run "cd ~" as soon as you open a new GitBash window.
  • Note: if you are opening a specific repo from GitHub Desktop you will NOT want to run this command, since you are already in the desired folder.