Python interpreter restart randomly takes ages

[pot-belly-mole]

I have mapped "Ctrl+Shift+F10" to "Interpreter: Restart active interpreter session" in Positron. I'm working on a qmd document with a python interpreter.

Sometimes when I trigger the restart, it is instantaneous, sometimes it takes maybe 5-20 seconds to do it. When it hangs, there is a little box in the right upper corner of the interpreter saying "Restarting", but the Python 3.13.9 exited (preparing for restart) appears only after the 5-20 seconds have passed. What might cause this seemingly random delay? Is it a bug?

I restart the interpreter very often while developing so this does become somewhat annoying.

Thank you for any insights in the matter!

Positron Version: 2025.10.1 build 4
Code - OSS Version: 1.103.0
Commit: b13dd1ca4803bc04a4a9165395b589b8caf4ab58
Date: 2025-10-15T00:40:08.583Z
Electron: 37.2.3
Chromium: 138.0.7204.100
Node.js: 22.17.0
V8: 13.8.500258-electron.0
OS: Linux x64 6.17.4-200.fc42.x86_64

[juliasilge]

That does sound really frustrating, and I also am a constant interpreter restarter! I can't say I've observed this so far, but we will definitely keep an eye on it. Would you say you've noticed anything that leads to it happening more? Like it happens more with uv vs. regular venv, or if your session had a lot of stuff in memory vs. not, or anything like that?

[pot-belly-mole]

Thank you for the reply!

At least I cannot seem to be able to reliably replicate the delay by creating a sizable amount of random data, say with:

import numpy as np
import os

def gen_data():
    return {os.urandom(20): np.random.randint(10)}

randomdata = {"data": [gen_data() for i in range(0, int(10e6))]}

The session's I've had it with definitely have not been bigger than that.

Happens with both uv and regular venv. And it does seem to mostly happen when I've had the interpreter session running for a longer time, eg. 15+ min.

[pot-belly-mole]

Will report if I notice some regularities in the behaviour!

[pot-belly-mole]

Here is some information I gathered from the console output channel during an incident where the restart hanged:

2025-11-05 16:53:23.196 [info] Restarting
2025-11-05 16:53:23.198 [warning] LSP startup timed out during interpreter restart
2025-11-05 16:53:23.198 [info] Stopping Positron LSP server, reason: restarting session
2025-11-05 16:53:38.528 [info] Restarting
2025-11-05 16:53:38.528 [warning] LSP startup timed out during interpreter restart
2025-11-05 16:54:34.355 [info] Kernel exited with code 0; waiting for restart to finish.
2025-11-05 16:54:34.356 [info] The kernel has started up after a restart.
2025-11-05 16:54:37.798 [info] Kernel is ready.
2025-11-05 16:54:37.799 [info] Set initial console width to 148
2025-11-05 16:54:37.802 [info] Starting Positron LSP server
2025-11-05 16:54:37.802 [info] Starting LSP server positron-lsp-python-9ead395c for 127.0.0.1
2025-11-05 16:54:37.829 [info] Starting Positron LSP client on port 40889

[juliasilge]

Are both of those "Restarting" messages from what you understand to be the same restart? i.e. the one that was hanging?

I do see the warning "LSP startup timed out during interpreter restart" one time during a healthy restart.

[pot-belly-mole]

Yes, both are the same restart, i.e. kernel restart was triggered only once for this, and you can see the time passing there in the timestamps. To me it seems like the first one never finishes and only after the second one is the kernel functional again.

[pot-belly-mole]

And another, no double this time (which makes me wonder if I had triggered the last one twice in frustration)

2025-11-06 16:33:58.056 [info] Restarting
2025-11-06 16:33:58.057 [warning] LSP startup timed out during interpreter restart
2025-11-06 16:33:58.057 [info] Stopping Positron LSP server, reason: restarting session
2025-11-06 16:34:15.127 [info] Kernel exited with code 0; waiting for restart to finish.
2025-11-06 16:34:15.127 [info] The kernel has started up after a restart.
2025-11-06 16:34:16.576 [info] Kernel is ready.

[juliasilge]

That is the same set of messages that I see for a healthy restart. Do you see anything different for a good, regular restart?

While you are looking at output channels, what do you see in "Python 3.X.X (name): Kernel" and "Python 3.X.X (name): Language Server (Console)"?

And if you are up for it, you could keep the Developer Tools console open for a while and see what happens there with a restart. I see this during a healthy restart:

INFO Restarting session 'Session Python 3.13.2 (Uv: scratch) (python-bdfb47ff) from runtime 88c2216f5999d4fd60fc0352d9a70016 (language: Python name: Python 3.13.2 (Uv: scratch) version: 3.13.2)' (Source: 'Restart Active Interpreter Session' command invoked)
INFO Creating positron.ui client 'positron-ui-python-18-a175d47d'...
INFO Creating positron.variables client 'positron-variables-python-19-f75a83f8'...
INFO Creating positron.help client 'positron-help-python-20-ea00f778'...