R - DEAD KERNEL - JupyterHub

962 views
Skip to first unread message

Matt Gardner

unread,
Aug 21, 2015, 8:13:49 PM8/21/15
to Project Jupyter
I have installed JupyterHub on a shared system with local users.  I have also configured the sudospawner following the great instructions here.  The install is on Ubuntu 14.04 running on an OpenStack cloud computing environment.  All seems to be working fine with the vanilla install.

My team can all log in to the server so my next task is to set up the R Kernel (later I will be adding Scala and Spark after I get R working). 

I installed the R kernel like this:

sudo apt-get install r-base-core
sudo apt-get install libzmq3-dev
sudo apt-get install python-zmq

Then from within R I ran:

> install.packages(c('rzmq','repr','IRkernel','IRdisplay'),
                  repos = c('http://irkernel.github.io/', getOption('repos')),
                  type = 'source')
> IRkernel::installspec(user=FALSE)


Once that ran I restarted the jupyterhub server and logged in - fantastic - I had R kernel working.....or so I thought.  When I asked my colleagues to sign in it looked good - they could see the R kernel (getting exciting now!) - but as soon as they launch an R notebook the R kernel crashed almost immediately with the message "DEAD KERNEL" - restarting did not help.  Any ideas where to start looking?

BTW - Python kernel works fine.

Matt






Matt Gardner

unread,
Aug 21, 2015, 8:27:59 PM8/21/15
to Project Jupyter
I should add that for me the RKernel works.  When I ran the install.packages in the previous post I started R as sudo.  The only difference is that first time I tried to install the kernel I ran the install.packages() in R under my username - which did not make the kernel available to other users on the JupyterHub.  I suspect that it is something to do with this.  I obviously don't want to have to run the install.packages for each user.

Matt

Michael Bright

unread,
Aug 23, 2015, 6:08:41 AM8/23/15
to Project Jupyter

Hi Matt,

Probably not your problem, but it might help others ...

I had a similar problem installing under Windoze 8 yesterday which was resolved by setting the path to R explicitly in the kernel.json file.

I'd installed the latest R, I followed those same steps as you to setup the IRKernel but then couldn't launch a notebook.

My problem was that I'd installed R but didn't have it in my PATH.

Probably less likely to be a problem with linux where binaries tend to go into standard places ... but worth a try.

Under Windows I have:
{"argv": ["c:/Progs/R/R-3.2.2/bin/R.exe", "--quiet", "-e","IRkernel::main()","--args",
"{connection_file}"],
"display_name":"R",
"language":"R"
}
 
Hope this helps someone,
Mike.

testb...@gmail.com

unread,
Apr 12, 2017, 6:46:22 AM4/12/17
to Project Jupyter

Hi Matt,
Could you please let me know how you got this issue resolved? I also have the same issue. Appreciate your help.
Thanks!
Reply all
Reply to author
Forward
0 new messages