Weird Problem with ParaFoam via SSH
 

Hi Guys,

I ran into some weird SSH/X11 issue when trying to run paraFoam through SSH from my Vista machine at home. For some reason paraFoam installed on my lab workstation (paraview-3.8.0 with OF-1.7.x) refuses to work though SSH showing the "segmentation fault" error message (it works on the workstation itself when I was in the lab earlier), while the paraFoam installed on the lab server (paraview-3.6.1 with OF-1.6.x) works fine through the SSH connection. I'm not sure if it's a problem with the x11 setting on my side or a badly installed paraview client on my work station, but for now I have to transfer all my files to the server (I'm running tests with OF-1.7 to see if I should switch everything over from 1.6) and postProcess from there. Any thoughts?

John

Greetings John,

Interesting... ParaView 3.6.1 works over ssh -X but 3.8.0 doesn't...
Personally I've always had problems with Putty+Xming... Xming tends to crash after its been working for a while, at least with me.

The only two possibilities that come to mind (using remote technology and without using Linux) and that should work flawlessly are:

1. If the ParaView you want to use is on your workstation, then try using a shared desktop connection, which should provide a full "mirror" of what you see on the workstation's desktop. You can also tunnel the VNC/Shared Desktop port through ssh.
2. Use the pvserver application that ParaView has (or at least ParaView 3.8.0 available at www.paraview.org has one) and then tunnel over ssh the port 11111 that pvserver uses. Instructions for pvserver available here: http://www.itk.org/Wiki/Setting_up_a_ParaView_Server
   Then use a local installation of ParaView 3.8.0 to plug to the pvserver's locally reflected port.

AFAIK, there is also another possible explanation, aside from the different ParaView versions: the graphic cards you are using in the workstation and in the lab server aren't the same type. For example:

• If your server's graphics card is Intel or something that doesn't even have 3D acceleration, than it is using the Mesa drivers, which rely mostly on CPU power to provide OpenGL graphics acceleration (which ParaView needs). This way, ParaView works well over ssh -X, because it doesn't require a local graphics card to do the OpenGL job.
• And if the workstation has a NVidia/ATI card with 3D acceleration and the whole-nine-yards of hardware based OpenGL, then you are stuck with requiring dedicated rendering on the spot. I.e., only locally can you use accelerated graphics performance.

By the way, did you test using ssh -X from the server into the workstation? Or at least from a computer in the lab that has Linux installed?

Best regards,
Bruno

Hi Bruno,

Thanks for the reply, it appears that the problem I'm having is caused by the second possibility that you suggested. The server was running with intel graphics while my work station was running on ATI HD 5xxx series, and the hardware was recent enough that I was having a hard time finding a working video driver for OpenSUSE 11.2. Hopefully the problem will go away after I upgrade my workstation to 11.3 later this week.

Thanks.

John