With all the latest jupyter and jupyterlab packages installed and a clean python environment, I can confirm that the jupyterlab UI becomes very slow if Show Contextual Help window is opened.
The issue also depends on the type of text/variable/function one types in the code cell; here are my observations:
If you are typing in a code cell a function which is already installed, for example, numpy.arange(…), the contextual help shows the corresponding help very quickly and thus no lag.
However, if you start typing in a code cell an object or variable which contains data, for example, a pandas dataframe say df, then as you type df, contextual help starts to load and show the data within that variable slowly. In addition, if you keep on typing like df.head, then the contextual help changes from the data content to the help contents of the function typed (in this case head()). This makes the whole jupyterlab UI freeze until the help is loaded.
I can confirm that reverting back to jupyter_client==7.0.6 solves the issue of having a cripplingly slow JupyterLab UI when several notebooks with large cell counts are open. Would be great to see this issue fixed at some point.
Painfully slow here; on both Chrome and Safari. Safari actually is worse and many NBExtensions won’t even work. Happened right after an OS upgrade was forced on me by corporate IT to MacOS Monterey 12.4. Tried downgrading; no change. Am not using Contextual Help.
From my observation upgrading to 3.4.xx above seems to introduce UI lagging issues, including delayed feedback when mouse hovering over any UI elements (including menu items, files on the folder browser, etc). Downgrading to 3.3.x (3.3.4 for my case) seems to remove the problem, at least on my setup. Possibly the performance issue could be caused by certain features that were introduced in between 3.3.4 → 3.4.X
Hello, please help. I have this issue for a long time and the lagging is really bad. I had version of 3.3.2 and lagging was still really bad for a big file. Now I upgrade it to 3.4.4 and it is still really bad lagging. I don’t have context help enabled. so could you please do something to fix it? I love jupyter lab but this lagging issue always make me thinking to move away from Jupyter lab. Please do something. Thank you so much!
Painfully slow here too. Please kindly fix this or assist us on fix this. A few month ago I have this issue, I went back to my old version. Now I created a new conda env and test if it is fixed, it is sill extremely slow. This is not updating works… Hope some issues can be located… and fixed.
Just thought I’d chime it that I have the same issue with the desktop app on Mac OSX 3.4.5-1 (seemingly, so take this with a grain of salt, it seems the recent update made it a bit worse)
for me, none of the above solutions works! I am so frustrated with jupyter lab, I love this program but it is too slow. could anyone know the other solution? Thanks
For anyone facing performance issues in 3.4, you can try to go into Settings → JupyterLab Shell → Hidden mode of main panel widgets → change it from scale to display (the change of default occurred in 3.4 and it seems to have negative impact on Chromium users - as discussed in #13092 - and therefore will be reverted in next point release)
try opera browser, for me jupyter lab works fast as light there. Maybe it is because I set up addblockplus in both firefox and chrome and they are full of cookies and cache.
Too slow, really too slow, too painfull here too: where a notebook opened before in less than one minute, it opens now in more than 9mn.
Linux Ubuntu 22 LTS
Chromium without any extension
via anaconda
jupyter_server 1.23.4
jupyter_client 7.4.9
jupyterlab 3.5.3
Don’t know if it is relevant, but I notice that in environnement, I have jupyter-server and jupyter-client python package installed (with a “-” in the name) in addition of jupyter_server and jupyter_client package (with a “_” in the name).
It is really hard to work on notebooks with this huge slow.
The problem is the same using a fresh mozilla firefox.