Kernel dies when using matplotlib to plot in notebook or save plot to image files

[zhu181]

Environment data

• VS Code version: 1.105.0
• Jupyter Extension version (available under the Extensions sidebar): 2025.9.0
• Python Extension version (available under the Extensions sidebar): 2025.16.0
• OS (Windows | Mac | Linux distro) and version: Windows_NT x64 10.0.26120
• Python and/or Anaconda version: 3.14.0
• Type of virtual environment used (N/A | venv | virtualenv | conda | ...): NA
• Jupyter server running: Local

Expected behaviour

Expect the matplotlib can show the image and save the image to file.

Actual behaviour

The kernel died.

Steps to reproduce:

[NOTE: Self-contained, minimal reproducing code samples are extremely helpful and will expedite addressing your issue]

1. Run the code.

import matplotlib.pyplot as plt
plt.plot([1, 2, 3], [4, 5, 6])

This bug is very close to Issue microsoft/vscode-jupyter#9283. I have been experimenting with it.

I found that this bug seems to be related to the python environments(ms-python.vscode-python-envs) extension. The bug only occurs when the extension is enabled and activated(in the python.useEnvironmentsExtension setting), and when using an interpreter that is not a conda or a venv environment interpreter(such as the pixi environment in this case.).

The bug would also sometimes occur even when using a conda environment interpreter if "python-envs.defaultEnvManager": "ms-python.python:system" is set in the vscode workspace.

I also experiment with other matplotlib backends(agg, qtagg, tkagg and so on) and other python versions(3.12-3.14). These do not affect the outcomes.

However the bug would not happen in the jupyter notebook running in the browser, or using the jupyterhub extension(ms-toolsai.jupyter-hub) in vscode to connect to a running jupyter server.

Logs

Output for Jupyter in the Output panel (View→Output, change the drop-down the upper-right of the Output panel to Jupyter)

18:05:32.718 [warn] Failed to get activated env vars for d:\Academic\global_soil_moisture_pattern\sm_pattern_new\.pixi\envs\default\python.exe in 30178ms
18:05:32.782 [info] Process Execution: d:\Academic\global_soil_moisture_pattern\sm_pattern_new\.pixi\envs\default\python.exe -c "import ipykernel; print(ipykernel.__version__); print("5dc3a68c-e34e-4080-9c3e-2a532b2ccb4d"); print(ipykernel.__file__)"
18:05:32.846 [info] Process Execution: d:\Academic\global_soil_moisture_pattern\sm_pattern_new\.pixi\envs\default\python.exe -Xfrozen_modules=off -m ipykernel_launcher --f=~\AppData\Roaming\jupyter\runtime\kernel-v324d4ded8293c9240fc9812540c55c92d907ccd53.json
    > cwd: d:\test
18:05:34.131 [info] Restarted cd4f7a9a-e0ff-4c9e-aab2-971daca8e25d
18:05:34.711 [error] Disposing session as kernel process died ExitCode: 3228369023, Reason: traitlets.traitlets.TraitError

[faultdiagnosistoolbox]

Edit:
After managing to dig deeper I found the problem I encountered and it was related to the update of ipykernel package to version 7.0.x. Downgrading to 6.30.1 resolves the issue.

Sorry for my interruption of this topic.

• VS Code version: 1.105.0
• Jupyter Extension version (available under the Extensions sidebar): 2025.9.1
• Python Extension version (available under the Extensions sidebar): 2025.16.0
• OS: Mac
• Python and/or Anaconda version: 3.12.09 and 3.13.3
• Type of virtual environment used (N/A | venv | virtualenv | conda | ...): venv
• Jupyter server running: Local

I think I have the same issue but I haven't been able to pinpoint exact versions of packages/extensions when it appeared.

The minimal example above excites the problem for me. If I run the code interactively with a Jupyter session, it plots nicely inline. But if I activate a matplotlib backend the session crashes when I plot (have tried the tk backend).

If I run directly in ipython; starts thebackend with %matplotlib and run the code it works as expected.

[DonJayamanne]

Thank you for the informaiton
@zhu181 Please could you try revering ipykernel as suggested.
@faultdiagnosistoolbox Do you run into similar isues outside vs code in Jupyter lab as well? If you do I suggest filing an issue upstream.

[faultdiagnosistoolbox]

Yes, tried it now and I experience exactly the same behavior in Jupyter lab.

I have reported the issue to ipykernel (I hope I got it in the right place in the issues system) ipython/ipykernel#1438 (comment)

[zhu181]

@DonJayamanne Reinstalling ipykernel doesn't fix this problem. Another minor issue is that the output panel doesn't always show kernel died reason. As the following shows

21:36:59.565 [info] Kernel successfully started
21:37:03.256 [error] Disposing session as kernel process died ExitCode: 3228369023, Reason: 

@faultdiagnosistoolbox I am not sure if the problem you encounter is related to this one, as I do not experience this behavior after your example in ipython/ipykernel#1438 (comment).

Also, downgrading ipykernel to 6.30 doesn't fix this problem either, so I consider this to be a problem with the python environment extension. Perhaps I should file an issue there?

[zhu181]

Move the code to a .py file and find the same problem: matplotlib fail to plot image when python environment extension is enabled. This seems more like a problem with the python environment extension.

[DonJayamanne]

matplotlib fail to plot image when python environment extension is enabled. This seems more like a problem with the python environment extension.

@zhu181 Do you run into this issue if you disable/uninstall the Python Environment extension?

[zhu181]

No, this issue will not occur if the Python Environment extension is disabled or uninstalled.

Edit:

Also, as I said, this issue will not occur if I set python.useEnvironmentsExtension: false.