Cell run to completion but kernel is still running - jupyter-notebook

I have a jupyter notebook does some data extractions. After executed the cell (no * with the cell) and get the extraction results, the kernel is still showing running (and CPU shows ipykernel 100%)
What could cause this happen and how to find out what process causes 100% usage on the ipykernel while no cell is running?

These normally caused by some extensions. Try to check if you have any extension for variable monitor/watch, then disable it.
Alternatively, you can try jupyter lab which has optimized many ways jupyter book issues with extensions.

Related

Jupyter notebook seems to remember previous path (!pwd) after being moved to a different directory?

I initially had a notebook in one directory in AWS SageMaker JupyterLab, say /A, but then moved it into /A/B. However, when I run !pwd in a jupyter notebook cell, I still get /A. This happens even when I press 'restart kernel'. How does the notebook remember this, and is there a way to prevent or reset this?
Thanks
I was actually using AWS SageMaker, and restarting the kernel from the toolbar was not enough. I needed to restart the kernel session, by pressing 'shut down' in the "Running terminals and kernels" section on the left navigation.
They are currently discussing warning users about the need to restart the kernel when a notebook is moved.

Using jupyter R kernel with visual studio code

For python jupyter notebooks I am currently using VSCode python extension. However I cannot find any way to use alternative kernels. I am interested in jupyter R kernel in particular.
Is there any way to work with jupyter notebooks using R kernel in VSCode?
Yes, it is possible. It just requires an additional configuration to connect with the R kernel in VSCode.
It's worth noting that, if you prefer, you can use the notebook in VSCode Insiders where there is native support for notebooks in many languages, including R.
If you're using Jupyter in VSCode, firstly install IRkernel (R kernel).
According to the docs, run both lines to perform the installation:
install.packages('IRkernel')
IRkernel::installspec() # to register the kernel in the current R installation
Now, you should:
Reload Window Ctrl + R
Type Ctrl + Shift + P to search for "Jupyter: Create New Blank Notebook"
Click on the button right below ellipsis in upper right corner to choose kernel
Switch to the desired kernel, in this case, R's
That's it!
Agreed with #essicolo, if you are 100% stuck on using vscode this is a no-go.
[About kernels] Sorry, but as of right now this feature is only supported with Python. We are looking at supporting other languages in the future.
Yeah, that's the case for now, even if you start an external server. I hate having to say that, as we really want to support more of the various language kernels. But we started out with a Python focus and we still are pretty locked into that for the near future. Polyglot support is coming, but it won't be right away
per Microsoft Employee IanMatthewHuff
https://github.com/microsoft/vscode-python/issues/5109#issuecomment-480097310
preface - based on the phrasing of your question, I am making the assumption that you are trying to perform IRkernel in-line execution from your text ide without having to use a jupyter notebook / jupyterlab.
That said, if you're willing to go to the dark side, there might be some alternatives:
nteract's Hydrogen kernel for Atom IDE - the only text ide that I'm aware of that still supports execution against IRkernel. I know, I know - it's not vscode but it's as close as you'll probably get for now.
TwoSigma's Beaker notebook - it's been a lonngggg time for me but this a branch of jupyter that used to support polyglot editing, I'm not sure if that's still supported and it seems like you aren't that interested in notebooks anyway.
#testing_22 it works with me too
just add some note from my experience
It will failed If you run IRkernel::installspec() from RStudio or from Jupyter Conda environment failed way
Please run this syntax with VSCode terminal
install.packages('IRkernel')
IRkernel::installspec()
The rest is same, please restart VSCode and select "R" kernel from VSCode

jupyter notebook takes forever to open and then pages unresponsive - [MathJax] issue

I'm trying to open a jupyter notebook and it takes a long time and I see at the bottom it's trying to load various [MathJax] extension, e.g. at the bottom left of the chrome browser it says:
Loading [MathJax]/extensions/safe.js
Eventually, the notebook loads, but it's frozen and then at the bottom left it keeps showing that it's trying to load other [MathJax] .js files.
Meanwhile, the "pages unresponsive do you want to kill them" pop up keeps popping up.
I have no equations or plots in my notebook so I can't understand what is going on. My notebook never did this before.
I googled this and some people said to delete the ipython checkpoints. Where would those be? I'm on Mac OS and using Anaconda.
conda install -c conda-forge nbstripout
nbstripout filename.ipynb. Make sure that there is no whitespace in the filename.
I had a feeling that the program in my Jupyter notebook was stuck trying to produce some output, so I restarted the kernel and cleared output and that seemed to do the trick!
If Jupyter crashes while opening the ipynb file, try "using nbstripout to clear output directly from the .ipynb file via command line"(bndwang). Install with pip install nbstripout
I was having the same problem with jupyter notebook. My recommendations to you are as follows:
First, check the size of the .ipynb file you are trying to open. Probably the file size is in MB and is large. One of the reasons for this might be the output of a dataset that you previously displayed all rows.
For example;
In order to check the dataset, sometimes I use pd.set_option('display.max_rows', None) instead of the .head() function. And so I view all the rows in the data set.
The large number of outputs increases the file size, making the notebook slower. Try to delete such outputs.
I think this will solve your problem.
Here restarting your kernel will not help. Instead use nbstripout to strip the output from command line.
Run this command -> nbstripout FILE.ipynb
Install nbstripout if it is not there
https://pypi.org/project/nbstripout/
It happened to me the time I decided to print a matrix for 100000 times. The notebook file became 150MB and Jupyter (in Chrome) was not able to open it: it said all the things you experienced and then the page died saying it was "OutOfMemory".
I solved the issue opening it in Visual Studio Code, there is a button "Clear All Output", then I saved the notebook again and it was back to some hundreds of KB, which I could open normally.
If you don't have Visual Studio Code installed, you can open the notebook with another editor (gedit if you use Linux or Notepad++ in Windows) and try to delete the output cells. This is more tricky since you have to pay a lot of attention in what you are deleting, otherwise the notebook will stop working.

IOPub data rate exceeded in Jupyter notebook (when viewing image)

I want to view an image in Jupyter notebook. It's a 9.9MB .png file.
from IPython.display import Image
Image(filename='path_to_image/image.png')
I get the below error:
IOPub data rate exceeded.
The notebook server will temporarily stop sending output
to the client in order to avoid crashing it.
A bit surprising and reported elsewhere.
Is this expected and is there a simple solution?
(Error msg suggests changing limit in --NotebookApp.iopub_data_rate_limit.)
Try this:
jupyter notebook --NotebookApp.iopub_data_rate_limit=1.0e10
Or this:
yourTerminal:prompt> jupyter notebook --NotebookApp.iopub_data_rate_limit=1.0e10
I ran into this using networkx and bokeh
This works for me in Windows 7 (taken from here):
To create a jupyter_notebook_config.py file, with all the defaults commented out, you can use the following command line:
$ jupyter notebook --generate-config
Open the file and search for c.NotebookApp.iopub_data_rate_limit
Comment out the line c.NotebookApp.iopub_data_rate_limit = 1000000 and change it to a higher default rate. l used c.NotebookApp.iopub_data_rate_limit = 10000000
This unforgiving default config is popping up in a lot of places. See git issues:
jupyter
IOPub data rate exceeded
It looks like it might get resolved with the 5.1 release
Update:
Jupyter notebook is now on release 5.2.2. This problem should have been resolved. Upgrade using conda or pip.
Removing print statements can also fix the problem.
Apart from loading images, this error also happens when your code is printing continuously at a high rate, which is causing the error "IOPub data rate exceeded". E.g. if you have a print statement in a for loop somewhere that is being called over 1000 times.
By typing 'jupyter notebook --NotebookApp.iopub_data_rate_limit=1.0e10' in Anaconda PowerShell or prompt, the Jupyter notebook will open with the new configuration. Try now to run your query.
Some additional advice for Windows(10) users:
If you are using Anaconda Prompt/PowerShell for the first time, type "Anaconda" in the search field of your Windows task bar and you will see the suggested software.
Make sure to open the Anaconda prompt as administrator.
Always navigate to your user directory or the directory with your Jupyter Notebook files first before running the command. Otherwise you might end up somewhere in your system files and be confused by an unfamiliar file tree.
The correct way to open Jupyter notebook with new data limit from the Anaconda Prompt on my own Windows 10 PC is:
(base) C:\Users\mobarget\Google Drive\Jupyter Notebook>jupyter notebook --NotebookApp.iopub_data_rate_limit=1.0e10
I have the same problem in my Jupyter NB on Win 10 when querying from a MySQL database.
Removing any print statements solved my problem.
For already running docker containers, try editing the file name - ~/.jupyter/jupyter_notebook_config.py
uncomment the line - NotebookApp.iopub_data_rate_limit =
and set high number like 1e10.
Restart the docker, it should fix the problem
I ran into this problem running version 6.3.0. When I tried the top rated solution by Merlin the powershell prompt notified me that iopub_data_rate_limit has moved from NotebookApp to ServerApp. The solution still worked but wanted to mention the variation, especially as internal handling of the config may become deprecated.
Easy workaround is to create a for loop and print. Then there wont be any issue. Printing directly wcc would cause if graph is huge. Hence any of below code will work as workaround.
wcc=list(nx.weakly_connected_components(train_graph))
for i in range(1,10):
print(wcc[i])
for i in wcc):
print(wcc)
Like others pointed out, print statement at a high rate can cause this. Resolve it by printing modulo a number using if statement. Example in python:
k = 10
if (i % k == 0):
print("Something")
Increase k if the warning persists.
Using Visual Studio Code, the Jupyter extension will be able to handle big data. launch from anaconda navigator
In general, trying to print something that is too long will trigger this error. I tried to print a string that was 9221593 characters long (too long), and that triggered the error.

IPython Notebook: Save the currently running notebook as ipynb file using a python command in a cell?

Is there are a way to save the Ipython notebook as an ipynb file from a cell within that notebook?
I know I can save it at any time by manually pressing "CTRL-M S", but I would like to use a command in a cell to do so (python command or %magic).
In this way I could "Run all cells" and be sure that the output (e.g. inline figures) is saved into the notebookfile when the execution is finished.
Update: Current versions of the Jupyter notebook (the successor of the IPython notebook) autosave into a hidden folder every few minutes (This feature was in development when I asked the question - see accepted answer).
No, because the kernel does not know it is accessed from a notebook. Dev version have auto-save feature though, and you could write a javascript extension that listen for cell execution event. But Python is not the way to do it. (or display(Javascript('js-code-to-save-notebook')) in the last cell, but I did not tell you)

Resources