Kitware / visualizer

The ParaViewWeb Visualizer application which can be used standalone or as a component within another web context.
https://kitware.github.io/visualizer/
BSD 3-Clause "New" or "Revised" License
115 stars 28 forks source link

Cannot connect to server in Open OnDemand #130

Open jasonleekungfu opened 2 hours ago

jasonleekungfu commented 2 hours ago

High-level description

I am trying to develop an Open OnDemand (OOD) app for ParaViewWeb (Visualizer, Divvy, Lite). I was able to create a Singularity image with ParaView 5.12.1 (headless) package and a working OOD app running this image. I was able to successfully submit a job on cluster, and see ParaViewWeb service correctly running. However, when I tried to connect to Visualizer, the "Loading ParaView" logo flashed and disappeared. Nothing happens. Similar issue also happened to Divvy and Lite, presumably somehow the app was not correctly talking to the service.

I tried ssh tunneling from local machine to the node and was able to successfully launch the app in browser by visiting "localhost". So it looks like the service is correctly running. But OOD was not able to talk to it correctly.

Steps to reproduce

This is the line to run the service once job starts (port is correct): singularity exec -B /work,/project,/usr/local/packages,/var/scratch /project/containers/images/paraviewweb-5.12.1.sif pvpython -m paraview.apps.visualizer --host 0.0.0.0 --port ${port} --timeout 0 -d

Detailed behavior

This is the output (debug) once a job starts:

image

When I tried to connect it in OOD, nothing changed in above output. As I described before, in browser, the "Loading ParaView" logo flashed and disappeared. The page is blank.

When I set up ssh tunneling and visited "localhost", the app was successfully launched. And I see below debug output:

image

So it looks like in OOD, it was unable to correctly talk to the service.

Expected behavior

OOD should be able to launch the app in browser.

Environment

jasonleekungfu commented 2 hours ago

Any help is appreciated. Thank you very much!