Brian Bonnlander (Mar 19 2020 at 14:25): Brian Bonnlander (Mar 19 2020 at 14:25):Hi everyone, JupyterHub appears down for Casper right now. Just checking if anyone is able to start JupyterLab on Casper DAV.
Anna-Lena Deppenmeier (Mar 19 2020 at 14:30):it works for me.
Brian Bonnlander (Mar 19 2020 at 14:31):Thanks! Are you using ssh-tunneling?
Anna-Lena Deppenmeier (Mar 19 2020 at 14:31):yes
Anna-Lena Deppenmeier (Mar 19 2020 at 14:32):using execdav and then start-jupyter
Brian Bonnlander (Mar 19 2020 at 14:39):I cannot connect to the browser when I ssh-tunnel. do not have "start-jupyter". Instead I am running:
(base) casper01:~$ execdav --account=STDD0003
Submitting interactive job to slurm ...
Setting display for X forwarding ...
Running command:
salloc --time 6:00:00 --partition dav --ntasks 1 --account=STDD0003 srun --pty --mem 0 --export=PATH=/bin:/usr/bin,SHELL,TERM,HOME,DISPLAY,XAUTHORITY /bin/bash -l
salloc: Pending job allocation 4590150
salloc: job 4590150 queued and waiting for resources
salloc: job 4590150 has been allocated resources
salloc: Granted job allocation 4590150
salloc: Waiting for resource configuration
salloc: Nodes casper10 are ready for job
Resetting modules to system default
(base) casper10:~$ jupyter lab --no-browser --ip=$(hostname)
[I 08:30:38.006 LabApp] [nb_conda_kernels] enabled, 7 kernels found
[I 08:30:40.904 LabApp] The port 8888 is already in use, trying another port.
[I 08:30:42.764 LabApp] [jupyter_nbextensions_configurator] enabled 0.4.1
[I 08:30:42.873 LabApp] JupyterLab extension loaded from /glade/u/home/bonnland/miniconda3/lib/python3.7/site-packages/jupyterlab
[I 08:30:42.873 LabApp] JupyterLab application directory is /glade/u/home/bonnland/miniconda3/share/jupyter/lab
[I 08:30:42.875 LabApp] Serving notebooks from local directory: /glade/u/home/bonnland
[I 08:30:42.875 LabApp] The Jupyter Notebook is running at:
[I 08:30:42.875 LabApp] http://casper10:8889/?token=dade741a8c9b67a8310943df559170eb36eb4531027876e5
[I 08:30:42.875 LabApp] or http://127.0.0.1:8889/?token=dade741a8c9b67a8310943df559170eb36eb4531027876e5
[I 08:30:42.875 LabApp] Use Control-C to stop this server and shut down all kernels (twice to skip confirmation).
[C 08:30:42.884 LabApp]
To access the notebook, open this file in a browser:
file:///glade/u/home/bonnland/.local/share/jupyter/runtime/nbserver-194264-open.html
Or copy and paste one of these URLs:
http://casper10:8889/?token=dade741a8c9b67a8310943df559170eb36eb4531027876e5
or http://127.0.0.1:8889/?token=dade741a8c9b67a8310943df559170eb36eb4531027876e5
Then I am giving this command in another local window:
ssh -N -L 8889:casper10:8889 casper10.ucar.edu
Does this look correct? Should my browser then be able to reach "http://casper10:8889/?token=dade741a8c9b67a8310943df559170eb36eb4531027876e5" ?
Michael Levy (Mar 19 2020 at 14:40):@Brian Bonnlander you should go to http://localhost:8889
Michael Levy (Mar 19 2020 at 14:40):(including the token string is fine)
Brian Bonnlander (Mar 19 2020 at 14:42):Thank you everyone, I try....
Michael Levy (Mar 19 2020 at 14:43):If that doesn't work, try changing the casper10.ucar.edu to casper01.ucar.edu in your ssh -N -L step.
Michael Levy (Mar 19 2020 at 14:44):I was thinking that the casper compute nodes are only accessible through the login nodes, but if you can ssh directly to casper10.ucar.edu then that is not the case
Brian Bonnlander (Mar 19 2020 at 14:45):Your suggestion of http://localhost:8889 works for me. Thank you again.
Brian Bonnlander (Mar 19 2020 at 15:06):One more question (hopefully): My ssh-tunnel session is working, but my Dask dashboard URL is apparently not working. It assigned a random port to me, but it seems problematic that "jupyterhub.ucar.edu/dav" is part of the URL. If http://jupyterhub.ucar.edu/dav is not resolving in my browser, should I expect my Dask dashboard URL to work? The full URL is "https://jupyterhub.ucar.edu/dav/user/bonnland/proxy/44116/status".
Deepak Cherian (Mar 19 2020 at 15:16):Use proxy/44116 in the dask lab extension
Brian Bonnlander (Mar 19 2020 at 15:17):Thank you! That works.
Brian Bonnlander (Mar 27 2020 at 19:07):Currently my attempts to run execdav are not working. Has anyone else had trouble with this in the past? Does this happen if you exceed your core hour allocation? This same command was working for me before.
(base) casper26:~$ execdav --account=STDD0003 Submitting interactive job to slurm ... No display; X forwarding disabled ... Running command: salloc --time 6:00:00 --partition dav --ntasks 1 --account=STDD0003 srun --pty --mem 0 --export=PATH=/bin:/usr/bin,SHELL,TERM,HOME /bin/bash -l salloc: error: Job submit/allocate failed: Invalid account or account/partition combination specified
Brian Bonnlander (Mar 27 2020 at 19:15):I see now that I forgot to ssh with X11 forwarding, but I still get the same error when using "ssh -Y" to login.
Brian Bonnlander (Mar 27 2020 at 19:31):OK, I finally worked my way around sam.ucar.edu and discovered that my core hour usage has gone over quota. So this is probably the source of the error message.
Last updated: May 16 2025 at 17:14 UTC