You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Recently I've noticed a behaviour that in my opinion should be corrected: if my laptop looses the internet connection and then I close the corresponding tab in the web browser - the kernel keeps running in the background.
To reproduce that behaviour one needs to do the following steps:
run a notebook in the app-mode.
disconnect the computer from the internet
close the tab
restore the internet connection
... After that the kernel keeps running "forever"
Similar thing happens if one often puts his computer into a "sleep" mode: once the python kernel gets detached from jupyter app - the only way to kill it is to go to the "taks" and shut it down manually.
I think there should be an automatic check that kills the kernel if it was not doing anything a certain amount of time. This will significantly reduce the use of computational resources.
The text was updated successfully, but these errors were encountered:
While I agree that zombie kernels are an issue, I don't think this should be address by Appmode.
In fact, the Jupyter developers have been discussing this for some time: ipython/ipython#5539 and apparently they recently came up with a solution: jupyter/notebook#2215 .
Recently I've noticed a behaviour that in my opinion should be corrected: if my laptop looses the internet connection and then I close the corresponding tab in the web browser - the kernel keeps running in the background.
To reproduce that behaviour one needs to do the following steps:
... After that the kernel keeps running "forever"
Similar thing happens if one often puts his computer into a "sleep" mode: once the python kernel gets detached from jupyter app - the only way to kill it is to go to the "taks" and shut it down manually.
I think there should be an automatic check that kills the kernel if it was not doing anything a certain amount of time. This will significantly reduce the use of computational resources.
The text was updated successfully, but these errors were encountered: