Hello everyone, I’m running JupyterLab via JupyterHub on a Linux server, but I’m encountering an issue now. I’ve been trying to solve it for three days without success, so I’m here to ask for your help.
My server system is on the computation server, but I deployed JupyterHub on the storage server. The storage server is mounted to the computation server via NFS. I created a Python virtual environment and installed all the required components in it. However, I ran into a problem when downloading SQLite—it failed to download, and I got stuck here. I could only download it to the /var or /opt directory of my computation server. So I used psycopg to configure c.JupyterHub.db_url. Now I can open Jupyter normally in the local browser, and I can import packages without issues. But when I try to read data, the kernel shows as idle, while there’s still an asterisk in front of the code block.
It’s worth noting that I previously installed and configured JupyterHub in the /opt directory of the computation server, and that worked perfectly. I don’t understand why so many problems arise when I move it to a different location. Does this mean JupyterHub can’t be installed on a mounted storage server?
Network storage can sometimes be a problem if it doesn’t have full POSIX support, or if it’s configured for particular types of workloads that don’t match what you’re trying to do.
If one of the problems you encountered was downloading sqlite (as opposed to running it) then I think your problems are at a much lower level than Jupyter. Maybe talk to your storage administrator?
Thank you very much for your reply. I regret to inform you that although I am currently pursuing my PhD, I have been appointed as a server administrator (with only experience in Linux operations and no experience in server architecture, etc.). I think my confusion regarding SQLite is that it can be deployed and runs perfectly normally on the compute server, but it cannot function properly on the mounted storage server. I noticed that the official website describes this as followsThe Hub’s Database — JupyterHub documentation:
Therefore, I switched to and configured a PostgreSQL database. Then, JupyterLab on the mounted storage server ran normally, but the issue mentioned earlier still occurred: loading packages works fine, but the kernel crashes as soon as I use the packages to read data. I don’t know which aspect to look into to solve this problem. Anyway, thank you very much for your help. I would be extremely grateful if you are willing to provide further assistance.
Earlier you said there was a problem downloading sqlite, not just using it. Together with the problems you’ve seen when trying to read files this suggests a problem beyond what Jupyter or Python can workaround.
You could try looking at the NFS client and server logs, and the kernel logs, for clues.
