I have recently got the access for Intel DevCloud as requested for completing the DC to Edge course.
I am successfully able to start the node using Putty but unfortunately not able to perform any task after qsub -I.
I am following the instructions as specified in the "Installation Instructions" doc which comes along the learning material, but at the first command itself which is "conda update conda" I am getting the following error : "EnvironmentNotWritableError: The current user does not have write permissions to the target environment." Moving further, if I try to create a new environment using the environment-win.yml file, the "Solving Environment" step encounters failure as it says the environment is inconsistent. I believe this particular step would work if the initial step works (conda update conda). Please suggest me a workaround for the same.
FYI, I can successfully pass through all the steps in my local machine but not on Intel DevCloud. I have attached a screen shot of the issue for a better understanding.
Thanks for reaching out to us.Kindly follow the below instructions.
1)Create environment with conda create --name env ,where env is your environment name.
2)Then activate your environment with conda activate env command.
3)After activating the environment,use conda update conda command.
Please let us know if the problem persists.
Thank you for the update, however I am still stuck at installing anaconda (conda install anaconda) in the new environment i.e tf_training. It is taking way longer than usual at the "Solving Environment" step, been over three hours now.
I have come across this issue on GitHub which conveys there are many other instances where conda install seems to take longer to install some packages in the environment. Is there any known workaround with you for this?
Thank you for the solution and I would want to let you know that the solution worked and that I am able to perform tasks now.
However, as the environment created in the steps you have provided doesn't load any dependencies, I am updating the environment using command "conda env update -f environment_win.yml" stated that current activated environment is "tf_training". And similar behaviour is observed i.e. "Solving Environment" fails with another error - ResolvePackageNotFound.
I am quite sure I am following the correct steps as installation happens properly in my local system but Intel DevCloud encounters this issue.
Thank you, that helped.
Now I am facing a similar issue raised on the forum here. Please have a look at the comment on 10/25/2019 02:25.
Error occurs at the LOC where cv2 is being imported. I have followed the instructions given by a coordinator from Intel on the topic link mentioned above, but that doesn't seem to work as well.
Attaching a screenshot for your reference.
Could you please let us know the following things:
1. Did all the packages installed successfully after the below command:
conda env create -f environment.yml
2. Please share the output screenshot of the following commands after activating the environment in the jupyter terminal.
You can find the terminal from the "New" dropdown at the right side of jupyter home page.
conda activate tf_training
3. In terminal type python and then give import cv2 and share the screenshot of it.
Following are my observations:
As per your initial instructions i have created the environment by "conda create --name tf_training" and then updated the yml file by typing in command: "conda env update -f environment_win.yml" stated that current activated environment is "tf_training". All the packages were installed successfully.
PFA required screenshots.
When can I expect a reply from your side? I have got the DevCloud access for a limited time (19 days left out of the permitted 30, to be precise) and I wish to make the most learning out of it during this tenure. I request you to please provide required support.
Kindly follow the below instructions:
To fix the issue in your Jupyter notebook:
> open jupyter terminal and give cd /home/u*****/.local/share/jupyter/kernels/tf_training
> Open kernel.json and check if the python path referenced is /glob python or path to your tf_training. If it is /glob python, change the path to /home/u*****/.conda/envs/tf_training/bin/python
> Open the terminal and activate tf_training. source activate tf_training
> Check the path of Python by giving the following command: which python. If it references /glob python path, update the PATH by exporting the conda python path.
i.e. export PATH="/home/u*****/.conda/envs/tf_training/bin":$PATH
> On checking which python, it should show the updated python PATH.
Now your issue should be sorted. Please verify and let us know if the solution provided worked. If the issue still persists, share the steps you have followed with the corresponding screenshots.
Note: Change uxxxx to your user id