Bug Manual

Bug Manual

Place to put common bugs + how to fix them (as a developer). PLEASE KEEP THIS PAGE UPDATED

Tips:

• Never modify files while running them
• Git CMD, Anaconda Prompt, CMD Prompt (Windows), GitKraken are helpful
• Any time you cannot spin up the frontend, a quick fix to try is delete the node_modules/ directory in ~/Deep-Learning-Playground/frontend/playground-frontend and then npm install followed by npm start *If you are unable to start the frontend or it fails immediately, you might have a 32 bit version of node. Install a 64 bit version to fix.

Troubleshooting:

1. Issue: Can’t delete /node_modules directory (Windows)

   Fix: Make sure that the Deep-Learning-Playground folder is outside of OneDrive. OneDrive has tendency to sync and that takes a while since /node_modules directory is HUGE!

2. Issue: Dev Server has been initialized using an options object that does not match the API Schema. -options.allowedHosts[0] should be a non-empty string

   Fix: In order, try: - Connecting to a non-GT WiFi like your home wifi or hotspot - If failed, revert what you did. Go into package.json in frontend/playground-frontend directory in the project, change react-scripts version from 5.* to 4.0.3. - If failed, revert what you did. cd into frontend/playground-frontend and run set DANGEROUSLY_DISABLE_HOST_CHECK=true && npm start to start the frontend If that doesn't work, revert the react-scripts version back to its original version. Then, remove the "proxy": "http://localhost:8000" line from package.json.

   Then, delete node_modules & package.lock files from the frontend/playground-frontend directory, run npm install followed by npm start and ensure that localhost:3000 pops up! See this link

3. Issue: Backend won’t start up because torchvision module doesn’t exist

   Fix: Run pip install torchvision in Anaconda Prompt while in the dlplayground conda environment. That should fix the issue!

4. Issue: Feedback form page doesn’t show up

   Fix: Install AWS CLI, configure the AWS credentials, cd into .aws, and run python build_env.py. Ask Faris if instructions are unclear

5. Issue: Pytest work properly on local environment, but github actions raise File Not Found Error

   Fix: Github actions run all pytests without entering the tests directory. Update file paths of tests folder or files in tests folder in a format similar to this - test_dir = '' if (os.getcwd()).split('\\')[-1].split('/')[-1] == 'tests' else 'tests'

6. Issue: version GLIBCXX_3.4.30' not found (required by /usr/share/miniconda/envs/dlplayground/lib/python3.9/site-packages/scipy/optimize/_highs/_highs_wrapper.cpython-39-x86_64-linux-gnu.so)`

   Fix: Make sure in the environment.yml file, scipy == 1.7.3 is set. Apparently, when fastai imports a submodule from sklearn, scipy version 1.9 gets imported, which doesn't seem to compatible with the gcc running on Github Actions.

Bug Manual

Place to put common bugs + how to fix them (as a developer). PLEASE KEEP THIS PAGE UPDATED

Tips:

• Never modify files while running them
• Git CMD, Anaconda Prompt, CMD Prompt (Windows), GitKraken are helpful
• Any time you cannot spin up the frontend, a quick fix to try is delete the node_modules/ directory in ~/Deep-Learning-Playground/frontend/playground-frontend and then npm install followed by npm start *If you are unable to start the frontend or it fails immediately, you might have a 32 bit version of node. Install a 64 bit version to fix.

Troubleshooting:

1. Issue: Can’t delete /node_modules directory (Windows)

   Fix: Make sure that the Deep-Learning-Playground folder is outside of OneDrive. OneDrive has tendency to sync and that takes a while since /node_modules directory is HUGE!

2. Issue: Dev Server has been initialized using an options object that does not match the API Schema. -options.allowedHosts[0] should be a non-empty string

   Fix: In order, try: - Connecting to a non-GT WiFi like your home wifi or hotspot - If failed, revert what you did. Go into package.json in frontend/playground-frontend directory in the project, change react-scripts version from 5.* to 4.0.3. - If failed, revert what you did. cd into frontend/playground-frontend and run set DANGEROUSLY_DISABLE_HOST_CHECK=true && npm start to start the frontend - If failed, revert what you did. Then, remove the "proxy": "http://localhost:8000" line from package.json. Then, delete node_modules & package.lock files from the frontend/playground-frontend directory, run npm install followed by npm start and ensure that localhost:3000 pops up! See this link

3. Issue: Backend won’t start up because torchvision module doesn’t exist

   Fix: Run pip install torchvision in Anaconda Prompt while in the dlplayground conda environment. That should fix the issue!

4. Issue: Feedback form page doesn’t show up

   Fix: Install AWS CLI, configure the AWS credentials, cd into .aws, and run python build_env.py. Ask Faris if instructions are unclear

5. Issue: Pytest work properly on local environment, but github actions raise File Not Found Error

   Fix: Github actions run all pytests without entering the tests directory. Update file paths of tests folder or files in tests folder in a format similar to this - test_dir = '' if (os.getcwd()).split('\\')[-1].split('/')[-1] == 'tests' else 'tests'

6. Issue: version GLIBCXX_3.4.30' not found (required by /usr/share/miniconda/envs/dlplayground/lib/python3.9/site-packages/scipy/optimize/_highs/_highs_wrapper.cpython-39-x86_64-linux-gnu.so)`

   Fix: Make sure in the environment.yml file, scipy == 1.7.3 is set. Apparently, when fastai imports a submodule from sklearn, scipy version 1.9 gets imported, which doesn't seem to compatible with the gcc running on Github Actions.

7. Issue: Boto3 Error: botocore.exceptions.NoCredentialsError: Unable to locate credentials

   Fix: Once you have AWS CLI installed (installation link in README; you can verify it has been installed by typing aws --version in the Terminal), type in aws configure and enter the following details as prompted. Not everything is required.

   User name	Password	AWS Access key ID	AWS Secret access key	Console login link	Default region name
   dev_user	DSGT_dlp123!	AKIAXS5IJR3CZNTV3A7L	5VC7xCxnbNYvbD15MO8vClVFBF8THYtxEeEk48Wp	https://521654603461.signin.aws.amazon.com/console	us-west-2