Using +NEW freezes Jupyterlab completely
michaelaye opened this issue · comments
Michael Aye commented
Describe the issue:
After pressing +NEW button to start a cluster:
- The launching never finishes, seemingly stalls.
- I cannot CTRL-C the Jupyter lab server in the terminal any more.
- I cannot open existing notebooks anymore.
- I do not get any log related to having tried to launch a dask cluster in the terminal.
- I still can get out of the tmux (the detach hotkey still works, so that i can kill the session that way.
As usual, this "WORKED BEFORE" (TM) on this machine, but I hadn't used it for several weeks. :(
Minimal Complete Verifiable Example:
- create a new env with python 3.11
- mamba install jupyterlab dask_labextension
- run
jupyter lab .
in a folder with work projects (sub folders with notebooks in them)
Anything else we need to know?:
I saw that a new py311 env automatically gets bokeh > 3, so I mamba-installed bokeh<3 to make dashboards work.
Environment:
- Dask version: 2023.1.0 (labextension 6.0.0)
- Python version: 3.11
- Operating System: Linux CentOS 7.5.1804 (core)
- Install method (conda, pip, source): conda
Michael Aye commented
Phew, found the issue. was caused by temp files redirected to non-existing node: dask/distributed#7491
Takaya Uchida commented
I think I'm getting a similar issue with the following errors showing up on Terminal:
[W 2023-02-25 22:10:57.992 ServerApp] wrote error: 'Cluster failed to start: Scheduler failed to start.'
[E 2023-02-25 22:10:57.996 ServerApp] {
"Host": "localhost:8000",
"Accept": "*/*",
"Referer": "http://localhost:8000/lab?",
"User-Agent": "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/110.0.0.0 Safari/537.36"
}
[E 2023-02-25 22:10:57.997 ServerApp] 500 PUT /dask/clusters?1677381055482 (127.0.0.1) 1691.83ms referer=http://localhost:8000/lab?
Takaya Uchida commented
The issue may have to do with the new version of Notebook..?