Restart not taken into account in v7

Questions and remarks about code_saturne usage
Forum rules
Please read the forum usage recommendations before posting.
daniele
Posts: 153
Joined: Wed Feb 01, 2017 11:42 am

Re: Restart not taken into account in v7

Post by daniele »

Hello Yvan,

Thank you for your analysis, I understand that you can't help more precisely without a more detailed debugging. And I don't exclude some simple stupid error from my side...
Anyway, I tried to uncheck the "use unmodified checkpoint...", but no improvement.
And yes, I double checked several times, the .csm file exists and seems ok, it makes 3GB.

The mesh makes 14M cells, it will be difficult to share it with you.
I attempted running CS with Valgrind: I attach the output, franckly speaking I don't have the knowledge to understand what's inside... Maybe it can help?

Thank you very much for your help and your time.
Best regards,
Daniele
Attachments
slurm-5511050.txt
(876.2 KiB) Downloaded 477 times
Yvan Fournier
Posts: 4153
Joined: Mon Feb 20, 2012 3:25 pm

Re: Restart not taken into account in v7

Post by Yvan Fournier »

Hello,

To run Valgrind, you need to run it on the cs_solver executable. There are instructions in the user guide for this, but it generally does not work well (or needs additional settings to avoid false positives) on a cluster with a fast network, and with a large mesh, it will be difficult to use in any case.

So the remaining option would be to attach a debugger. I can provide some instructions, or maybe this can be taken offline in a separate discussion (I believe you have my contact info).

Best regards,

Yvan
StandardRANSUser001
Posts: 16
Joined: Tue Apr 09, 2024 3:26 am

Re: Restart not taken into account in v7

Post by StandardRANSUser001 »

Hi Yvon and Daniele,

I am running CS on a HPC, and I would like to restart my cases without the GUI. The case is built similar to Daniele's where I have a setup.xml file, and I am now including cs_user_script.py in the DATA folder. The python file is being added to the RESU folder when the case compiles/runs, but the case is restarting at t = 0, rather than at the last checkpoint that I have nominated in cs_user_scripts.py.

I have included the following lines in the define_domain_parameters function:
domain.mesh_input = 'RESU/20240614-1653/mesh_input'
domain.restart_input = 'RESU/20240614-1653/checkpoint'

The remander of the cs_user_script.py is the defult from when the case was created with the --copy-ref flag.

The RESU/20240614-1653/checkpoint folder contains auxiliary.csc, main.csc, mesh_input.csm.

It's unclear to me whether more case information should be included in this python script. Any guidance on this would be greatly appreciated.

Best regards,
Sean Hanrahan
Yvan Fournier
Posts: 4153
Joined: Mon Feb 20, 2012 3:25 pm

Re: Restart not taken into account in v7

Post by Yvan Fournier »

Hello,

I don't use the cs_user_scripts.py for this very often, and for restarting without the GUI, the "auto restart' (which automatically chooses the restart with the latest timestamp) is more practical, but before checking in more detail, can you simply try :

domain.restart_input = 'RESU/20240614-1653'

Best regards,

Yvan
daniele
Posts: 153
Joined: Wed Feb 01, 2017 11:42 am

Re: Restart not taken into account in v7

Post by daniele »

Hello,

Sorry I missed your post.
My personal way of using cs_user_scripts.py to restart a calculation, is just to add:
domain.restart_input = 'RESU/"case_name"/checkpoint'.
Leaving all the other lines "None".

Let me know if it works.
Best regards,
Daniele
Post Reply