Notebook: jupyter kernal keeps dying

[jslug1000]

Hi,

When opening any jupyter notebook, even one with no code in, I get the 'Dead kernal' message:

The kernel has died, and the automatic restart has failed. It is possible the kernel cannot be restarted. If you are not able to restart the kernel, you will still be able to save the notebook, but running code will no longer work until the notebook is reopened.

My terminal window shows the following:

[I 07:56:47.888 NotebookApp] [nb_conda_kernels] enabled, 1 kernels found
[I 07:56:48.230 NotebookApp] The port 8888 is already in use, trying another port.
[I 07:56:48.438 NotebookApp] [nb_anacondacloud] enabled
[I 07:56:48.456 NotebookApp] [nb_conda] enabled
[I 07:56:48.874 NotebookApp] ✓ nbpresent HTML export ENABLED
[W 07:56:48.874 NotebookApp] ✗ nbpresent PDF export DISABLED: No module named 'nbbrowserpdf'
[I 07:56:48.886 NotebookApp] Serving notebooks from local directory: /Users/jedgecombe
[I 07:56:48.886 NotebookApp] 0 active kernels
[I 07:56:48.886 NotebookApp] The Jupyter Notebook is running at: http://localhost:8889/
[I 07:56:48.887 NotebookApp] Use Control-C to stop this server and shut down all kernels (twice to skip confirmation).
[W 07:57:08.143 NotebookApp] Notebook Dropbox/Python/AB_testing/template/ab_test_template.ipynb is not trusted
[I 07:57:08.694 NotebookApp] Kernel started: 8743e528-acb5-488f-8e46-456675500373
[W 07:57:08.723 NotebookApp] 404 GET /kernelspecs/Python%20[Root]/logo-64x64.png (::1) 25.64ms referer=http://localhost:8889/notebooks/Dropbox/Python/AB_testing/template/ab_test_template.ipynb
Traceback (most recent call last):
File "/Users/jedgecombe/anaconda3/lib/python3.5/runpy.py", line 184, in _run_module_as_main
"__main__", mod_spec)
File "/Users/jedgecombe/anaconda3/lib/python3.5/runpy.py", line 85, in _run_code
exec(code, run_globals)
File "/Users/jedgecombe/anaconda3/lib/python3.5/site-packages/ipykernel/__main__.py", line 2, in
from ipykernel import kernelapp as app
File "/Users/jedgecombe/anaconda3/lib/python3.5/site-packages/ipykernel/kernelapp.py", line 98, in
ConnectionFileMixin):
TypeError: metaclass conflict: the metaclass of a derived class must be a (non-strict) subclass of the metaclasses of all its bases
[I 07:57:11.695 NotebookApp] KernelRestarter: restarting kernel (1/5)
Traceback (most recent call last):
File "/Users/jedgecombe/anaconda3/lib/python3.5/runpy.py", line 184, in _run_module_as_main
"__main__", mod_spec)
File "/Users/jedgecombe/anaconda3/lib/python3.5/runpy.py", line 85, in _run_code
exec(code, run_globals)
File "/Users/jedgecombe/anaconda3/lib/python3.5/site-packages/ipykernel/__main__.py", line 2, in
from ipykernel import kernelapp as app
File "/Users/jedgecombe/anaconda3/lib/python3.5/site-packages/ipykernel/kernelapp.py", line 98, in
ConnectionFileMixin):
TypeError: metaclass conflict: the metaclass of a derived class must be a (non-strict) subclass of the metaclasses of all its bases

It tries to restart 5 times (so some repeated code from above) then shows the following:

[W 07:57:23.735 NotebookApp] KernelRestarter: restart failed
[W 07:57:23.735 NotebookApp] Kernel 8743e528-acb5-488f-8e46-456675500373 died, removing from map.
ERROR:root:kernel 8743e528-acb5-488f-8e46-456675500373 restarted failed!
[W 07:57:23.757 NotebookApp] Kernel deleted before session
[W 07:57:23.757 NotebookApp] 410 DELETE /api/sessions/d129aa7e-fc50-4cfd-9e5d-aab57ff7b49e (::1) 1.74ms referer=http://localhost:8889/notebooks/Dropbox/Python/AB_testing/template/ab_test_template.ipynb

I've tried pip uninstalling and re-installing, also tried this with conda but no luck and can't see similar problems online. Would be extremely grateful for any help!

[takluyver]

Can you try uninstalling all of:

• ipykernel
• ipython
• jupyter_client
• jupyter_core
• traitlets
• ipython_genutils

And then install again. If you're doing this inside a conda env, it might be easiest to create a new environment and start from scratch. Also, if you're going to install with conda, run conda clean -tipsy to clean up conda caches before you start.

[jslug1000]

fixed it! Thank you so much - I've been searching for hours for an answer! Really appreciate it.

[takluyver]

You're welcome. For the record, my guess is that it was incompatible versions of some of those packages being installed. It probably wasn't necessary to reinstall them all, but it's easier to do that than try to work it out more closely.

[sainttelant]

thanks so much, i 've tried several solutions, but didn't work until i found uninstall ipykernel, and reinstalled it, it worked,thanks.

[jonnyhtw]

Brilliant, worked straight away.

Cheers

Jonny

[Dan-Git-source]

I am trying to work with jupyter notebook, but when I open a file I receive the following error:

The kernel has died, and the automatic restart has failed. It is possible the kernel cannot be restarted. If you are not able to restart the kernel, you will still be able to save the notebook, but running code will no longer work until the notebook is reopened.

In the CMD I see the following:

(base) C:UsersDan Eran>jupyter notebook [W 19:05:33.006 NotebookApp] Error loading server extension jupyterlab Traceback (most recent call last): File "C:UsersDan EranAppDataRoamingPythonPython36site-packagesnotebooknotebookapp.py", line 1451, in init_server_extensions mod = importlib.import_module(modulename) File "C:Anaconda3libimportlib__init__.py", line 126, in import_module return _bootstrap._gcd_import(name[level:], package, level) File "", line 994, in _gcd_import File "", line 971, in _find_and_load File "", line 953, in _find_and_load_unlocked ModuleNotFoundError: No module named 'jupyterlab' [I 19:05:33.122 NotebookApp] Serving notebooks from local directory: C:UsersDan Eran [I 19:05:33.122 NotebookApp] 0 active kernels [I 19:05:33.122 NotebookApp] The Jupyter Notebook is running at: [I 19:05:33.122 NotebookApp] http://localhost:8888/?token=99a355c23c6617857e387f53d0af607ae26b89c20598336e [I 19:05:33.122 NotebookApp] Use Control-C to stop this server and shut down all kernels (twice to skip confirmation). [C 19:05:33.122 NotebookApp]

Copy/paste this URL into your browser when you connect for the first time,
to login with a token:
http://localhost:8888/?token=99a355c23c6617857e387f53d0af607ae26b89c20598336e
[I 19:05:33.247 NotebookApp] Accepting one-time-token-authenticated connection from ::1 [I 19:05:42.699 NotebookApp] Creating new notebook in [I 19:05:43.563 NotebookApp] Kernel started: 1433cbbf-f4b9-4dd3-be19-e91d7ee3d82f Traceback (most recent call last): File "C:Anaconda3librunpy.py", line 193, in _run_module_as_main "main", mod_spec) File "C:Anaconda3librunpy.py", line 85, in _run_code exec(code, run_globals) File "C:Anaconda3libsite-packagesipykernel_launcher.py", line 15, in from ipykernel import kernelapp as app File "C:Anaconda3libsite-packagesipykernel__init__.py", line 2, in from .connect import * File "C:Anaconda3libsite-packagesipykernelconnect.py", line 13, in from IPython.core.profiledir import ProfileDir ModuleNotFoundError: No module named 'IPython' [I 19:05:46.555 NotebookApp] KernelRestarter: restarting kernel (1/5), new random ports Traceback (most recent call last): File "C:Anaconda3librunpy.py", line 193, in _run_module_as_main "main", mod_spec) File "C:Anaconda3librunpy.py", line 85, in _run_code exec(code, run_globals) File "C:Anaconda3libsite-packagesipykernel_launcher.py", line 15, in from ipykernel import kernelapp as app File "C:Anaconda3libsite-packagesipykernel__init__.py", line 2, in from .connect import * File "C:Anaconda3libsite-packagesipykernelconnect.py", line 13, in from IPython.core.profiledir import ProfileDir ModuleNotFoundError: No module named 'IPython' [I 19:05:49.591 NotebookApp] KernelRestarter: restarting kernel (2/5), new random ports Traceback (most recent call last): File "C:Anaconda3librunpy.py", line 193, in _run_module_as_main "main", mod_spec) File "C:Anaconda3librunpy.py", line 85, in _run_code exec(code, run_globals) File "C:Anaconda3libsite-packagesipykernel_launcher.py", line 15, in from ipykernel import kernelapp as app File "C:Anaconda3libsite-packagesipykernel__init__.py", line 2, in from .connect import * File "C:Anaconda3libsite-packagesipykernelconnect.py", line 13, in from IPython.core.profiledir import ProfileDir ModuleNotFoundError: No module named 'IPython' [I 19:05:52.620 NotebookApp] KernelRestarter: restarting kernel (3/5), new random ports Traceback (most recent call last): File "C:Anaconda3librunpy.py", line 193, in _run_module_as_main "main", mod_spec) File "C:Anaconda3librunpy.py", line 85, in _run_code exec(code, run_globals) File "C:Anaconda3libsite-packagesipykernel_launcher.py", line 15, in from ipykernel import kernelapp as app File "C:Anaconda3libsite-packagesipykernel__init__.py", line 2, in from .connect import * File "C:Anaconda3libsite-packagesipykernelconnect.py", line 13, in from IPython.core.profiledir import ProfileDir ModuleNotFoundError: No module named 'IPython' [W 19:05:53.595 NotebookApp] Timeout waiting for kernel_info reply from 1433cbbf-f4b9-4dd3-be19-e91d7ee3d82f [I 19:05:55.632 NotebookApp] KernelRestarter: restarting kernel (4/5), new random ports WARNING:root:kernel 1433cbbf-f4b9-4dd3-be19-e91d7ee3d82f restarted Traceback (most recent call last): File "C:Anaconda3librunpy.py", line 193, in _run_module_as_main "main", mod_spec) File "C:Anaconda3librunpy.py", line 85, in _run_code exec(code, run_globals) File "C:Anaconda3libsite-packagesipykernel_launcher.py", line 15, in from ipykernel import kernelapp as app File "C:Anaconda3libsite-packagesipykernel__init__.py", line 2, in from .connect import * File "C:Anaconda3libsite-packagesipykernelconnect.py", line 13, in from IPython.core.profiledir import ProfileDir ModuleNotFoundError: No module named 'IPython' [W 19:05:58.671 NotebookApp] KernelRestarter: restart failed [W 19:05:58.671 NotebookApp] Kernel 1433cbbf-f4b9-4dd3-be19-e91d7ee3d82f died, removing from map. ERROR:root:kernel 1433cbbf-f4b9-4dd3-be19-e91d7ee3d82f restarted failed! [W 19:05:58.705 NotebookApp] 410 DELETE /api/sessions/fd456273-adb3-48cd-92f8-d531c9b8f7a8 (::1): Kernel deleted before session [W 19:05:58.709 NotebookApp] Kernel deleted before session [W 19:05:58.709 NotebookApp] 410 DELETE /api/sessions/fd456273-adb3-48cd-92f8-d531c9b8f7a8 (::1) 4.00ms referer=http://localhost:8888/notebooks/Untitled11.ipynb?kernel_name=python3

I have tried to uninstall and then reinstall the modules you've mentioned. However, I was not able to solve the problem. any ideas? THANKS!!

[ShixiangWang]

I can not fix this problem 😢

[paulmattheww]

It must be a keras specific issue, tensorflow alone works fine on my machine but keras keeps crashing it.

[alvi-subhan]

i am having this issue whenever i try to run keras or tensorflow it keeps crashing while its working perfectly with other libraries.

[gunjannaik]

Remove the ipykernel using conda remove ipykernel
and then resinstall with lower version with pip install ipykernel ==4.2.2

Worked for me.

[alvi-subhan]

thankyou so much :)

On Wed, May 30, 2018, 4:16 PM Gunjan Naik [e-mail] wrote:

Remove the ipykernl using conda remove ipykernel
and then resinstall with lower version with pip install ipykernel ==4.2.2

Worked for me.

—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
https://github.com/jupyter/notebook/issues/1892#issuecomment-393123918,
or mute the thread
https://github.com/notifications/unsubscribe-auth/AhsClxJV9FajcyIJhQXvMsj5mSne2rXpks5t3n-qgaJpZM4KxC7D
.

[fazzalll]

if anyone facing the issue regarding .. kernel failure

it may be probably due to ipython failure

you can fix it by reinstalling ipython and its dependencies on your machine
$ pip install ipython[all]
you can see kernel working properly in jupyter

it worked for me

[thetrailblazer4]

if you're using anaconda then run this in anaconda prompt "conda clean -tipsy" then conda install the package. for e.g. if you're working with Numpy package, then run conda install numpy

[Sakshi2018]

In my case, When I tried running the cells, they were working fine untill reached to the training cell, where the kernel showed an error:

[Sakshi2018]

However, the jupyter notebook looks like:

Any help will be greatly appreciated

[park-moonkyu]

thanks a lot.
i have a problem about dead kernel
but it solved

[AliceDinh]

I got that problem, I first conda remove ipykernel then pip install ipykernel==4.2.2
I got errors: ModuleNotFoundError: No module named 'traitlets'
I install 'traitlets' by: conda install traitlets successfully
but then I type "jupyter notebook"
ModuleNotFoundError: No module named 'traitlets'

[gunjannaik]

I guess Conda installed libraries are not accessible to your ipykernel (which is installed using pip).

Install it using pip install traitlets.

[AliceDinh]

I did try with pip install traitlets but still raising that error

[AliceDinh]

Hi,
I followed again the instruction made by @takluyver and it's working now, thanks all

[BishnoiAmit]

Thanks @takluyver. Just reinstalling the ipykernel did the trick.

I think what @takluyver is trying to say in his above post is that, try to reinstall these libraries one by one and hopefully it'll resolve the problem (don't just panic when you see a bunch of libraries re-installation tackle them one by one).

And I would personally suggest to go with the ipykernel first.

Other than that, if you're getting an error like this: failed to create cublas handle: CUBLAS_STATUS_ALLOC_FAILED

Then try allowing memory growth:

config = tf.ConfigProto()
config.gpu_options.allow_growth = True
session = tf.Session(config=config)

Taken from link (https://github.com/tensorflow/tensorflow/issues/7072#issuecomment-306061314). Commented by @shingte

Cheers!

[Gravypan]

I've tried every fix recommended here and none of them have worked.

[joeltorby]

I got the same issue. Tried all of this but none worked. Trying to run the python code in jupyter notebook gave a new error "The kernel appears to have died. It will restart automatically". By googling this error I found a new solution:

pip install tensorflow==1.5

I've tried to install other versions of tensorflow before, but they didn't do the trick. Also make sure to have the latest version of numpy.

pip install --upgrade numpy

[duhaime]

To augment @takluyver's helpful note:

pip uninstall ipykernel ipython jupyter_client jupyter_core traitlets ipython_genutils -y
pip install ipykernel ipython jupyter_client jupyter_core traitlets ipython_genutils

[AzimAhmadzadeh]

thanks so much, i 've tried several solutions, but didn't work until i found uninstall ipykernel, and reinstalled it, it worked,thanks.

Only uninstalling/installing this single library (ipykernel) seems to be enough.

[nataliele]

Thanks, I tried updating ipykernel and it works again now!

[alfredolopez80]

Hello, Guy, i solve the problem with this command

sudo -H pip3 install fs
sudo -H pip3 install prompt-toolkit --force-reinstall
pip3 install prompt-toolkit --force-reinstall

[altayg]

in case it helps someone, my similar problem is solved by running the followings that I had seen somewhere else:

on Ubuntu terminal:

$ jupyter kernelspec list

$ rm -r /home/me/.local/share/jupyter/kernels/python3

$ python -m ipykernel install --user

[adi5krish]

Hello,
I have tried every fix recommended and none of them solved the issue. Any help would be appreciated.
Thanks in advance

[kevin-bates]

Hi @adi5krish - Since these kinds of issues can manifest themselves in a million different ways, it might be helpful to get specific information. Could you please provide the following and perhaps others will have some helpful ideas? Thanks.

1. What python (version, conda, etc.) and operating system?
2. The output of jupyter kernelspec list.
3. Reproduced with --debug enabled.
4. The complete console output of the Notebook server with --debug enabled.

[adi5krish]

Hii kevin,
The python version i am using is 3.6.9 and the operating system i am running in is Windows 10
This is the output i am getting when i run jupyter kernelspec list is
Available kernels:
python3 C:UsersapulikonAppDataRoamingjupyterkernelspython3
I am sorry, i couldn't understand what exactly you meant about reproduced with --debug enabled.
I am sharing the error i am getting when i try to run the code

Let me know if you need anything else

[kevin-bates]

I'm sorry, I should have been more clear. If you start notebook as follows: jupyter notebook --debug, it will produce debug output. That said, I doubt we're going to see much more in the console, but it never hurts.

It looks like your kernel knows something about tensorflow and gpus. Is it getting this "idea" based on the execution of a cell? Or is something baked into your argv stanza of the kernelspec? I suspect the former, which would imply the kernel is starting up, but then some action (presumably the cell that just executed) is causing the kernel process to die. Notebook server then detects that the kernel process is no longer running and enters its auto-restart sequence.

Unless others have more direct advice, I suggest you try to narrow things down by starting from a base configuration and adding things (like tensorflow, then gpu, etc.) until the kernel dies.

[BHARATBHUSHAN9923]

Don't know what is happening before it was just fine. I tried a reinstall of anaconda but didn't helped.

[berkai]

I've the same issue. In the command prompt I saw the ImportError: cannot import name 'create_prompt_application' issue after that just ran pip3 install ipython than worked for me.

[LACoderDeBH]

conda clean -tipsy by itself worked for me, thank you!

[awais0005]

i am having this issue whenever i try to run keras or tensorflow it keeps crashing while its working perfectly with other libraries.

did you get the solution?
if yes please revert back to me.

[maulberto3]

I tried to continue working on an Ipython in another PC, but even after doing pipenv's, I still get the kernel died issue.

what's more, I uploaded the said IPython file to Google Colab, and it's still dying inside there too.

Very strange.

[eyildiz-ugoe]

Getting the same error. Not using conda. Tried everything above but no luck. Any further idea?

[may-phyu]

Now I face the problem of dead kernel in jupyter notebook when I reach 43-epoch because there is 50-epoch in my code. It took two nights to reach 43-epoch by using two GPUs. I run it two times but it didn't work. I want to get the last epoch. Please advise me what I need to do.

[kevin-bates]

Hi @may-phyu - this sounds very frustrating. Unfortunately, there isn't much to go on here. You'll need to figure out why the kernel is dying after 43 epochs.

• Is the kernel out of resources?
• In both runs, has it died after the 43rd epoch? If so, is there anything special about 43 (besides being one more than 42 - the answer to Life, the universe and everything! [Sorry, bad time for that.])?
• Do you have access to other resources that better support your training run?
• How does the notebook (kernel) come into play relative to the epochs? Notebooks are typically interactive - perhaps converting the code into a straight python application is better suited for this situation at this point?
• Is the kernel running local to the Notebook server?
• Is the Notebook server/kernel running in a container environment?

Should you decide to run your current experiment again via Notebook, I would restart the Notebook server with the --debug flag and capture the command console's output (redirect to a file, etc.). Then, assuming the kernel dies again, include that command console output (all of it!) in a response.

Good luck.

[SaurabhTripathi62]

conda uninstall ipykernel

them conda install ipykernel

solved my resolve

[nani-samireddy]

igot this error please help me

[furqan4545]

@LACoderDeBH

conda clean -tipsy by itself worked for me, thank you!

Your so called command removed all my updated anaconda packages. I wish I could kill you.

[nani-samireddy]

in my case
it is saying that

[nani-samireddy]

can anyone help me
plzzzzzzzzzzzzzz

[furqan4545]

Re install The following

• ipykernel
• ipython
• jupyter_client
• jupyter_core
• traitlets
• ipython_genutils

On Thu, Mar 26, 2020, 1:22 PM nani samireddy [e-mail]
wrote:

can anyone help me
plzzzzzzzzzzzzzz

—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
https://github.com/jupyter/notebook/issues/1892#issuecomment-604293404,
or unsubscribe
https://github.com/notifications/unsubscribe-auth/AL7UYYBJRYLCVRDJ3CC4P23RJMGEDANCNFSM4CWEF3BQ
.

[nani-samireddy]

do you mean uninstalling and installing
??????

[furqan4545]

Yess exactly. And try working in a conda environmemt

On Thu, Mar 26, 2020, 1:31 PM nani samireddy [e-mail]
wrote:

do you mean uninstalling and installing
??????

—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
https://github.com/jupyter/notebook/issues/1892#issuecomment-604297368,
or unsubscribe
https://github.com/notifications/unsubscribe-auth/AL7UYYFLNFDTDF5XQEMNNLLRJMHHBANCNFSM4CWEF3BQ
.

[SaurabhTripathi62]

bro make a new env then reinstall it works

[gqlo]

Tried all the options didn't work. Ran notebook with Python3 worked for me.
python3 -m notebook

[Ashishtyagi-smsd]

https://stackoverflow.com/questions/59576397/python-kernel-dies-on-jupyter-notebook-with-tensorflow-2/62658027#62658027
I have mentioned the solution that worked for me with tensorflow 2.

[vasunagpal1]

Can you try uninstalling all of:

• ipykernel
• ipython
• jupyter_client
• jupyter_core
• traitlets
• ipython_genutils

And then install again. If you're doing this inside a conda env, it might be easiest to create a new environment and start from scratch. Also, if you're going to install with conda, run conda clean -tipsy to clean up conda caches before you start.

Where Do I exactly run this?

[Pavithra2001]

Hello
I am also facing the same issue. The kernel in jupyter notebook is not starting.
I tried upgrading all packages and also uninstalling and reinstalling all those mentioned above. But that too didn't work for me. Is there any other method to solve this?
Thank You!

[Jyothsna131]

I am also facing an issue of dead kernel. All cells are running except the cell that contains model.fit() function(training).
Details:
Windows 10
ipykernel version 5.3.4
python 3.7.9
tensorflow 2.1.0
keras 2.3.1
Please suggest me how to solve this.

[kevin-bates]

Hi @Jyothsna131 - please see https://github.com/jupyter/notebook/issues/1892#issuecomment-585802775

I suspect you're running out of resources and that terminates the kernel process. Is there any output produced in the notebook or in the console window in which you started the notebook server?

You might try monitoring resources (CPU/Memory) prior to the kernel process crashing.

Does your notebook code run to completion outside of notebook (i.e., via python directly) in that same environment?

[Jyothsna131]

@ Kevin-bates
Maybe that's the reason, sir. Thank you so much for your reply.

[harry7171]

Hi ,

I am using GCP compute engine for running jupyter notebook with 8 cores and 32 GB RAM
I am running Randimized Search CV for XGBoost with around (139999,754) data for training
I have been trying to fix this since long. Any Help would be greatful. :)