CEA Codes forum

Hi,

I'm having issues when running error studies on a linux computing cluster.

These are the steps I follow:
1) I connect to a computing node that has reserved 50 cores for my job.
2) In the list of computers in the 'Errors'-tab I just have the Local (0.0.0.0) computer selected with 50 cores.
3) I start the error study.

Now from time to time the study finishes without any issues, but most of the time I get the following segmentation fault:

*** Error in `TraceWin_noX11': double free or corruption (out): 0x00007f029c0049e0 ***
...
Critical system error: Abnormal termination (abort).

It seems like a memory-related issue, but how to circumvent this?
When I run the same routine with the GUI ('TraceWin'), I don't have this problem.

Kind regards,

Lennert

Dear Lennert,

I've carried out a few similar tests on my own without observing this type of error. So it's probably due to your configuration that I don't know about, or some options in your projects. Could you send it to me if it's not too complicated ?
But I've noticed display problems with the Linux version, which I've corrected and put on the site.

Regards,

Didier

The files are added in the attachment (field maps exluded). Cheers, Lennert

Dear Lennert,

In similar configurations I've never had any problems, so maybe it's down to your project, but I can't test it without the field map files!

Regards,

Didier

You can find the field maps in the attachment. Thanks in advance! Kr, Lennert

Dear Lennert,

I've fixed the bug that caused the noX11 version to crash, but the GUI version is normally affected too, although it seems more stable.

Regards,

Didier

Cool, thanks! Kr, Lennert