restarting the Python Interactive kernel increases persistent memory usage

[rbavery]

Environment data

• VS Code version: 1.37
• Extension version (available under the Extensions sidebar): 2019.8.30787
• OS and version: Ubuntu 18.04
• Python version (& distribution if applicable, e.g. Anaconda): conda-forge 3.7.3
• Type of virtual environment used (N/A | venv | virtualenv | conda | ...): conda
• Relevant/affected Python packages and their versions: XXX
• 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): XXX

Expected behaviour

Wouldn't persistently increase memory use

Actual behaviour

When I restart the kernel in a Python Interactive session, my memory use goes up and stays up. If I restart a bunch of times in a row, it keeps going up. Only way to clear is to close vscode entirely. It goes up by about .08 Gb with each restart, which adds up quickly.

Steps to reproduce:

1. start an interactive session, star restarting the kernel, and monitor the memory with htop

Logs

There was nothing in the logs

Output from Console under the Developer Tools panel (toggle Developer Tools on under Help; turn on source maps to make any tracebacks be useful by running Enable source map support for extension debugging)

see attached, couldn't copy paste from dev tools screen. I've also included a gif showing the problem.

[IanMatthewHuff]

@rbavery. Thanks for the report on this. We made a change recently to our restart to always start up a second session under the hood to allow for a fast restarts. But it looks like on restart there are python processes building up and not being disposed of. I'll look into this.

[tigerhawkvok]

@IanMatthewHuff I can confirm this on my system -- I have a very memory intensive process that I wanted to dump, and on restarting the kernel it didn't dispose of the old process. I needed to manually kill the process.

Version: 1.37.1 (system setup)
Commit: f06011ac164ae4dc8e753a3fe7f9549844d15e35
Date: 2019-08-15T16:17:55.855Z
Electron: 4.2.7
Chrome: 69.0.3497.128
Node.js: 10.11.0
V8: 6.9.427.31-electron.0
OS: Windows_NT x64 10.0.17763

Name: Python
Id: ms-python.python
Description: Linting, Debugging (multi-threaded, remote), Intellisense, code formatting, refactoring, unit tests, snippets, and more.
Version: 2019.8.30787
Publisher: Microsoft
VS Marketplace Link: https://marketplace.visualstudio.com/items?itemName=ms-python.python

[mgsnuno]

I also confirm exactly the same issue on my system. I noticed the restarts getting faster, which is nice, but then I noticed a build up of processes.

Easy to reproduce:

1. Start IPython interactive in VS Code
2. run in Terminal pidof python
3. restart IPython kernel x times
4. run in Terminal pidof python and see x more processes

This has quite an impact while using dask for example, as the old cluster is not killed and the resources used by it are just stuck there.