Some common Jupyterhub issues and solutions
500: Internal Server Error
This can occur if you've installed software using pip at the user-level.
Check which version of nbconvert is currently installed as your user:
> pip list | grep nbconvert
You'll want a version >= 7.2.10
> pip install --upgrade --user nbconvert
----
Got 'xxxxx' instead of '0o0600'
The full error message will typically identify a specific file. For example:
.local/share/jupyter/runtime/kernel-1086e611-42ed-476e-92a7-d5be14bcda67.json'.
Got '0o0' instead of '0o0600'.
In this case, the expected file permissions do not match actual permissions. Locate the file and change the permissions:
> chmod 0600 kernel-xxxxx.json
You may need to relaunch JHub.
R: Dead Kernel
This is a documented issue related to png(), X11, and R
The workaround is to create a file in your home directory named .Rprofile
.
By default all new accounts will have this file automatically created in their home directory. Double-check that this file exists with the following contents:
options(bitmapType='cairo')
Kernels won't load
Take a look at the permissions of the following subdirectory in your home directory:
~username/.local/shared/jupyter/runtime
runtime should have 3700 permissions:
> stat runtime/
File: runtime/
Size: 32768 Blocks: 64 IO Block: 262144 directory
Device: 3eh/62d Inode: 100877607 Links: 2
Access: (3700/drwx--S--T) Uid: (10336/ petkus) Gid: (10010/ itstaff)
If it does not, change the permissions:
> chmod 3700 ~/.local/shared/jupyter/runtime
Robert Petkus, 6/7/23