Troubleshooting jupyter notebooks session not starting

If you have reached your quota in your $HOME you will not be able to start a jupyter notebook session (because the process is not going to be able to write needed files in your $HOME directory).

To check you quota you can use:




$ myquota

to check what is taking up space in $HOME you can run:

$ find $HOME -maxdepth 1 | xargs -i du -sh {} | sort -rh

This command will take a few minutes to run. You can also open a new terminal and run:




$ cd $HOME
$ du -h

The "find" command will give a neat summary of the top directories in $HOME that are taking up space, the "du" command will show all the files in your $HOME.

To check what is taking up space in the current directory you can use:




$ find ./ -maxdepth 1 | xargs -i du -sh {} | sort -rh

("./" indicated the current directory). The find commands will take some time to run depending on how much data you have in your local directory.

To alleviate the space issue you should remove or move directories (targeting the largest ones as much as possible - see output of the first "find" command above).

If you use anaconda heavily the following two directories could be the likely "culprit" taking up a large fraction of your $HOME space:

$HOME/.conda/envs

$HOME/.conda/pkgs

If the two above directories are taking up a significant fraction of your 40GB in $HOME and if (and only if) your group has purchased storage you could consider moving this directories there and there create a symbolic link of them back in their original location.

For example if your username is "joebruin" and your group is called "gobruins" and you have access to a directory such as:

/u/project/gobruins/joebruin

you could:

$ mkdir /u/project/gobruins/joebruin/dot.conda

$ mv $HOME/.conda/envs /u/project/gobruins/joebruin/dot.conda

$ mv $HOME/.conda/pkgs /u/project/gobruins/joebruin/dot.conda

NOTE: the two "mv" commands above may take a few minutes to complete it is important that you let them run to completion.

and then we sym-linked the directories back to their original position via:

$ ln -s /u/project/gobruins/joebruin/dot.conda/envs $HOME/.conda
$ ln -s /u/project/gobruins/joebruin/dot.conda/pkgs $HOME/.conda

NOTE: in lie of the sybmolic link you cound tell conda where to locate the directory via "conda config"

so that conda will still find them.

jupyter jupyter nb jupyter notebooks on h2

Creation date: 4/13/2023 4:54 PM Updated: 5/1/2024 1:32 PM