Jupyter Lab中conda环境关联Kernel调用错误Python解释器求助

已尝试现有类似问题的解决方案但均无效。从激活的conda环境snowflakes中安装Kernel后，Jupyter Lab仍调用系统Python解释器而非环境内的解释器，具体操作及现象如下：

操作流程 #

1. 激活conda环境

conda activate /opt/miniconda3/envs/snowflakes 

2. 安装关联环境的Kernel

python -m ipykernel install --user --name snowflakes --display-name snowflakes_2 

3. 启动Jupyter Lab

jupyter-lab

4. 在Notebook中查看Python路径

import sys
print(sys.path)

实际输出（指向系统Python）：

['/Users/user/Documents/Code/Python /PyCharm_Test',
 '/Library/Frameworks/Python.framework/Versions/3.10/lib/python310.zip',
 '/Library/Frameworks/Python.framework/Versions/3.10/lib/python3.10',
 '/Library/Frameworks/Python.framework/Versions/3.10/lib/python3.10/lib-dynload',
 '',
 '/Library/Frameworks/Python.framework/Versions/3.10/lib/python3.10/site-packages']

5. 预期输出（指向conda环境）：

['/Users/user/Documents/Code/Python /PyCharm_Test/src',
'/Users/user/Documents/Code/Python /PyCharm_Test',
'/opt/miniconda3/envs/snowflakes/lib/python310.zip',
'/opt/miniconda3/envs/snowflakes/lib/python3.10',
'/opt/miniconda3/envs/snowflakes/lib/python3.10/lib-dynload',
'/opt/miniconda3/envs/snowflakes/lib/python3.10/site-packages']

已尝试的无效操作：

• 多次重装ipykernel和Jupyter Lab
• 更换多种Kernel安装方式
• 通过Anaconda Navigator启动Jupyter Lab

解决方案 #

1. 手动修正Kernel配置文件 #

Jupyter Kernel的配置文件位于~/.local/share/jupyter/kernels/snowflakes/kernel.json，打开该文件，确认argv中的Python路径是否指向conda环境的解释器：

{
 "argv": [
  "/opt/miniconda3/envs/snowflakes/bin/python",
  "-m",
  "ipykernel_launcher",
  "-f",
  "{connection_file}"
 ],
 "display_name": "snowflakes_2",
 "language": "python"
}

若路径错误，手动修改为/opt/miniconda3/envs/snowflakes/bin/python，保存后重启Jupyter Lab。

2. 在conda环境内安装并启动Jupyter Lab #

避免系统级Jupyter调用错误解释器，直接在目标环境内部署Jupyter：

conda activate /opt/miniconda3/envs/snowflakes
conda install jupyterlab ipykernel -y
python -m ipykernel install --user --name snowflakes --display-name snowflakes_2
jupyter-lab

此方式启动的Jupyter会默认绑定当前conda环境的Python。

3. 清理旧Kernel配置后重新安装 #

若存在同名Kernel的缓存配置，先卸载再重装：

jupyter kernelspec uninstall snowflakes -y
conda activate /opt/miniconda3/envs/snowflakes
python -m ipykernel install --user --name snowflakes --display-name snowflakes_2

4. 确认conda环境激活状态 #

安装Kernel和启动Jupyter前，务必确认环境已正确激活：

conda activate /opt/miniconda3/envs/snowflakes
which python

输出需为/opt/miniconda3/envs/snowflakes/bin/python，若不是，执行source ~/.bashrc（或source ~/.zshrc）初始化conda后再重试。

内容的提问来源于stack exchange，提问作者Ludwig B