Error with Catalyst postprocessing

Questions and remarks about code_saturne usage
Forum rules
Please read the forum usage recommendations before posting.
Post Reply
Luciano Garelli
Posts: 243
Joined: Fri Dec 04, 2015 1:42 pm

Error with Catalyst postprocessing

Post by Luciano Garelli »

Hello,

I'm using CS 6-beta compiled with paraview 5.6 in order to use catalyst for postprocessing. I create the python script in paraview and set the writer in CS.

When I run the simulation using catalyst I get the following error.

Code: Select all

File "/share/apps/paraview/5.6.0/lib/python2.7/site-packages/paraview/coprocessing.py", line 178, in LoadRequestedData
datadescription.GetInputDescriptionByName(key).AllFieldsOn()
AttributeError: 'NoneType' object has no attribute 'AllFieldsOn'
I attach the paraview script.

Thanks in advance.

Regards,

Luciano
Attachments
catalyst_test.py
(9.7 KiB) Downloaded 37 times
Yvan Fournier
Posts: 3105
Joined: Mon Feb 20, 2012 3:25 pm

Re: Error with Catalyst postprocessing

Post by Yvan Fournier »

Hello Luciano,

Due to changes in ParaView's Catalys export itself, our documentation/tutorials is obsolete.

In versions prior to ParaView 5.5, the input read from a file was always named "input" in the Catalyst Python script. In current versions, the name is based on the file (in your case, it is named "RESULTS.case").

When the names do not match, you get the error you report. Due to the interaction between the Python and C parts, I did not find a way to make error reporting clearer than this...

You have the choice between 2 solutions:
  • In the generated script, replace "RESULTS.case" everywhere (2 or 3 occurences) with the name of the Catalyst writer ("catalyst" here)
  • Before exporting the script, in the top of the ParaView Pipeline, right-click the top entry (named RESULTS.case here, possibly named differently depending on the input file name), and rename it to the name of the writer ("catalyst" in this case)
This should get things working again.

Best regards,

Yvan
Luciano Garelli
Posts: 243
Joined: Fri Dec 04, 2015 1:42 pm

Re: Error with Catalyst postprocessing

Post by Luciano Garelli »

Hello Yvan,

Thank you for the information. I was reading the catalyst user guide and it is mentioned that the input "should be named “input”, the default value".
In paraview 5.6 has to used the "Generated script-deprecated" to set the input name, is the new methodology is used "Export Catalyst Script", the input name has to be change directly in the script file.

Now I'm facing some openGL problems and trying to solve it.

Thanks again for your help.

Regards,

Luciano
Yvan Fournier
Posts: 3105
Joined: Mon Feb 20, 2012 3:25 pm

Re: Error with Catalyst postprocessing

Post by Yvan Fournier »

Hello Luciano,

Which OpenGl error messages do you have, and which OpenGL library are you linking with ? Depending on the issue it may be something we have encountered and solved/worked around.

Best regards,

Yvan
Luciano Garelli
Posts: 243
Joined: Fri Dec 04, 2015 1:42 pm

Re: Error with Catalyst postprocessing

Post by Luciano Garelli »

Hello Yvan,

The problem with the OpenGL was with the off screen rendering of paraview, our cluster-admin fix the problem and now is working correctly, also the live visualitation.

Thanks for your help.

Regards,

Luciano.
Luciano Garelli
Posts: 243
Joined: Fri Dec 04, 2015 1:42 pm

Re: Error with Catalyst postprocessing

Post by Luciano Garelli »

Hello Yvan,

I'm running a case where a joining process between meshes is used. When I change the default results writer from MED to catalyst a get the following error.

Code: Select all

SIGSEGV signal (forbidden memory area access) intercepted!

Call stack:
   1: 0x7f71c95740a1 <+0xa20a1>                       (libsaturne-6.0.so)
   2: 0x7f71c9575605 <fvm_writer_export_nodal+0x15>   (libsaturne-6.0.so)
   3: 0x7f71c98d34ca <cs_join_post_mesh+0x5ea>        (libsaturne-6.0.so)
   4: 0x7f71c98d389e <cs_join_post_faces_subset+0x5e> (libsaturne-6.0.so)
   5: 0x7f71c98e1b64 <cs_join_split_faces+0x3a74>     (libsaturne-6.0.so)
   6: 0x7f71c98a3128 <cs_join_all+0x2088>             (libsaturne-6.0.so)
   7: 0x7f71c96421ea <cs_preprocess_mesh+0x35a>       (libsaturne-6.0.so)
   8: 0x7f71cb019892 <cs_run+0x1c2>                   (libcs_solver-6.0.so)
   9: 0x7f71cb019567 <main+0x167>                     (libcs_solver-6.0.so)
  10: 0x304801ed1d <__libc_start_main+0xfd>         (libc.so.6)
  11: 0x40430d     <>                               (cs_solver)
End of stack
I set the visualization of the joinning process to 0. If I restore the result writer to MED and create a new writer for catalyst the error dosen't appear and a joinning.med file with 0 KB is writed in the postprocessing directory.

I don't know if it is a bug or the result writer is allways needed.

Regards,

Luciano
Yvan Fournier
Posts: 3105
Joined: Mon Feb 20, 2012 3:25 pm

Re: Error with Catalyst postprocessing

Post by Yvan Fournier »

Hello Luciano,

This seems to be a bug, so I'll check. I seem to recall there may be an issue using the same Catalyst writer for preprocessing and "main" visualization due to the way meshes/inputs are named and used, but at the least, there should be a clean exit/and/or error message.

Thanks for the report.

Best regards,

Yvan
Luciano Garelli
Posts: 243
Joined: Fri Dec 04, 2015 1:42 pm

Re: Error with Catalyst postprocessing

Post by Luciano Garelli »

Hello Yvan,

One more question about Catalyst. It is possible to update the python script during the simulation, or once the simulation start the script is read only at the initialization of the writer?

Thanks.

Regards,
Luciano
Yvan Fournier
Posts: 3105
Joined: Mon Feb 20, 2012 3:25 pm

Re: Error with Catalyst postprocessing

Post by Yvan Fournier »

Hello Luciano,

In the writer, the script is read at initialization, so unless VTK does some tracking under the hood (which would surprise me), you can assume you cannot update it.

One possibility would be to modify the script so that t reads updates (this is full Python, so you can "eval" Python files read).

Best regards,

Yvan
Post Reply