You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The design of Pluto is very nice and cool! I've watched a few of the JuliaCon and other videos about it. Thanks for creating this!
My colleagues and I do a lot of collaborative data science work. Many of us use Jupyter Lab running on an HPC system, so we have direct compute access to the same data files, etc. Our university runs a "portal", where a web app helps users set up a compute allocation, then starts a Jupyter Lab session people can connect to.
But I like to use a newer Jupyter version, with my own stuff available (via singularity image, etc.). So I've figured out a way to do that via shell access, and the proxy web server (ngnix, I believe) scheme they use for the portal (still needs/uses university authentication).
Then I can access my custom server via their url scheme: https://sub.domain.edu/node/${host}/${port}/
On an HPC node, where I have Julia available, I can launch . . . julia> Pluto.run(;launch_browser=false, host="0.0.0.0")
There's not much to video, but this is all I see in my client workstation browser window:
I think (hope) that means I've actually connected to the Pluto web server, but the route being sent is not known (When I try other things, the error page looks more like nginx style).
I suspect Pluto is getting sent the node/${host}/${port}/ as part of the request (which it would obviously not know what to do with).
With my working Jupyter Lab setup, I have to include the following in my config file: c.ServerApp.base_url = '/node/${host}/${port}/'
I think that is necessary, because otherwise the Jupyter Lab server will always get that node/${host}/${port}/ as part of the request, and won't know what to do with that (those paths make no sense to it's web url API). I assume that setting just lops that part off, before sending to the Jupyter server.
Is there a similar setting available/possible for Pluto.run()?
Thanks!
P.S. I've confirmed with a local Pluto session that "Not found!" looks identical (same dark background, etc.), when I insert a nonsensical string after the port, in the browser url.
So I have some hope that I am actually able to get to Pluto server running on my HPC node, but maybe just need a way to remove that "base url".
The text was updated successfully, but these errors were encountered:
Thanks for the investigation! There are currently no way to remove the base url but this would definititely interesting to support directly inside Pluto.
The design of Pluto is very nice and cool! I've watched a few of the JuliaCon and other videos about it. Thanks for creating this!
My colleagues and I do a lot of collaborative data science work. Many of us use Jupyter Lab running on an HPC system, so we have direct compute access to the same data files, etc. Our university runs a "portal", where a web app helps users set up a compute allocation, then starts a Jupyter Lab session people can connect to.
But I like to use a newer Jupyter version, with my own stuff available (via singularity image, etc.). So I've figured out a way to do that via shell access, and the proxy web server (ngnix, I believe) scheme they use for the portal (still needs/uses university authentication).
Then I can access my custom server via their url scheme:
https://sub.domain.edu/node/${host}/${port}/
On an HPC node, where I have Julia available, I can launch . . .
julia> Pluto.run(;launch_browser=false, host="0.0.0.0")
There's not much to video, but this is all I see in my client workstation browser window:
I think (hope) that means I've actually connected to the Pluto web server, but the route being sent is not known (When I try other things, the error page looks more like nginx style).
I suspect Pluto is getting sent the
node/${host}/${port}/
as part of the request (which it would obviously not know what to do with).With my working Jupyter Lab setup, I have to include the following in my config file:
c.ServerApp.base_url = '/node/${host}/${port}/'
I think that is necessary, because otherwise the Jupyter Lab server will always get that
node/${host}/${port}/
as part of the request, and won't know what to do with that (those paths make no sense to it's web url API). I assume that setting just lops that part off, before sending to the Jupyter server.Is there a similar setting available/possible for Pluto.run()?
Thanks!
P.S. I've confirmed with a local Pluto session that "Not found!" looks identical (same dark background, etc.), when I insert a nonsensical string after the port, in the browser url.
So I have some hope that I am actually able to get to Pluto server running on my HPC node, but maybe just need a way to remove that "base url".
The text was updated successfully, but these errors were encountered: