Issue to be resolved
Frequent disconnects from JupyterHub with one of these messages:
- “Server unavailable or unreachable. Would you like to restart it?”
- “Kernel restarting: The kernel appears to have died. It will restart automatically.”
Solution
“Server unavailable or unreachable. Would you like to restart it?”
This message is exclusively related to intermittent network/connectivity issues between your client and the JupyterHub server. You may simply refresh the page to connect to an active JupyterHub session or start a new session if the session timeout has been reached. If you are onsite at GSFC and are regularly experiencing network issues, please contact the Enterprise Service Desk (ESD) to report those network problems:
https://esd.nasa.gov/esdportal
These prompts may also occur if your authentication through NASA’s Access Launchpad expires. Again, simply refresh the page or start a new session. Launchpad timeouts are outside of NCCS control and most frequently occur during JupyterHub sessions that last longer than 8 or 12 hours.
“Kernel restarting: The kernel appears to have died. It will restart automatically.”
This could also be caused by intermittent network connectivity, but usually it indicates either an out of memory issue or a software fault. You may need to either request a new session with more memory or investigate reducing the memory you use. You can also check your /home directory for logs. For example, “jhub-XXXX.out” on Discover or “slurm-XXXX.out” on ADAPT. These may offer a clue as to why the kernel is dying.
Please also consider submitting a batch job via Slurm for any long-running work. See the Slurm documentation below:
Discover: https://www.nccs.nasa.gov/nccs-users/instructional/using-slurm
ADAPT: https://www.nccs.nasa.gov/nccs-users/instructional/adapt-instructional/s…