Skip to content

Remote Jupyter connection with long running command: Extension host terminated due to memory  #3324

@simplelife2010

Description

@simplelife2010

Environment data

  • VS Code version: 1.35.1
  • Extension version (available under the Extensions sidebar): 2019.5.18875
  • OS and version: macOS Mojave 10.14.5
  • Python version (& distribution if applicable, e.g. Anaconda): 3.6.8 (Anaconda)
  • Type of virtual environment used (N/A | venv | virtualenv | conda | ...): conda
  • Relevant/affected Python packages and their versions: JupyterLab 0.35.6
  • Jedi or Language Server? (i.e. what is "python.jediEnabled" set to; more info How to update the language server to the latest stable version vscode-python#3977): True

Expected behaviour

When connecting to a remote Jupyter server, I expect the VS Code extension host not to terminate unexpectedly, even when I run long running commands remotely (> 50 minutes), for example training machine learning models using Tensorflow.

Actual behaviour

When connecting to a remote Jupyter server and executing long running (> 50 minutes) commands like model.fit(), after approx. 50 minutes I receive a popup 'Extension host terminated unexpectedly' with the option to restart the extension host. The connection to the Jupyter session is lost and I receive no further output from the long running command, even if I restart the extension host. I cannot reconnect to the kernel which seems to be still running (as I can see on the server). I also cannot interrupt or restart this kernel, as the corresponding buttons show no reaction. To continue working I have to kill the running kernel on the Jupyter server. When I have done this and try to start another Jupyter session from VS Code, I receive the message 'Cannot execute code, session has been disposed.'

Steps to reproduce:

  1. Setup a python file with ipython cells.
  2. Configure an external Jupyter server (specify Jupyter server URI)
  3. Execute a command via IPython running longer than 50 minutes.

Logs

No output in the output panel

Metadata

Metadata

Assignees

Labels

No labels
No labels

Type

No type
No fields configured for issues without a type.

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions