open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello all,

I have an issue with a mesh not partitioning that I cannot figure out. The error simple message suggests that the problem is on a solid boundary (uncomplicated island boundary). I can run the same Telemac config/compilation with other meshes (so I don’t think it’s the cas file or compilation issues) and this mesh will run on one processor but I cannot partition it at all. Can anyone help please? What am I missing? Although the channels aren’t refined in the most ideal way it usually runs like this and works OK for my application – is that the problem though?

I’m running T2D v8.2, have tried with CHECKING THE MESH on and off, and various different number of SC partitioning nodes. I attached the partel_T2DGEO log file, cas file, GEO and CLI files.

Many thanks in advance if anyone can give me some pointers.
Sophie

---

BOUNDARY 5 :
BEGINS AT BOUNDARY POINT: 8680 , WITH GLOBAL NUMBER: 219219
AND COORDINATES: 431256.0 5907868.
ENDS AT BOUNDARY POINT: 8680 , WITH GLOBAL NUMBER: 219219
AND COORDINATES: 431256.0 5907868.
THE MESH PARTITIONING STEP STARTS
BEGIN PARTITIONING WITH METIS
RUNTIME OF METIS 0.0000000E+00 SECONDS
THE MESH PARTITIONING STEP HAS FINISHED
ERROR IN BOUNDARIES J= 5598
WHILE NPTFR_P(I)= 5577
WHILE L= 5577



PLANTE: PROGRAM STOPPED AFTER AN ERROR
RETURNING EXIT CODE: 2

geo files is missing and *.log file is unreadable.
Please take time to read the rules to join the files (join a zip of the files)
Regards

Hello Christophe. Apologies, that's my bad, the log ought to have been converted to a txt. I've converted it, zipped it all (<5MB) and cannot now post with the zip attached - I assume there is some sort of size limit to attachments which is why the slf didn't load the first time? I can't see that in the rules tab though, sorry. I could try some sort of other file share option?

Hi
It seems to me that size is limited to 2MB
Create separate zip in separate posts
Or give a link to a download place where people could access....

WeTransfer link to files: we.tl/t-NejI8dH9ud

Hope that works now, it's the SLF which is big so cannot be sent even in an individual post. You don't need an account to access the files, they're available there for 7 days if anyone does get a change to take a look...Thanks!

Hello slr,

I would suggest you to read the following post:
www.opentelemac.org/index.php/kunena/16-...ry-in-parallel#21414

CHECKING THE MESH = YES tells you that you have 338 overconstrained triangles which means your mesh is low quality and should be improved.

It seems that this may be the issue when running in parallel.

So my advice is to change your mesh to delete these overconstrained triangles (3 nodes belonging to boundaries). Moreover, your simulation should be better without such elements.

Hope this helps,

Chi-Tuan