MPI Error in REEF3D
 

Hi there,

I have been trying to run a 3D VIV simulation using the latest version of REEF3D. I created a new directory and moved REEF3D and DIVEMesh along with the ctrl.txt and control.txt.

Upon running the sim, the following error comes up:-

driver FSI initialize
[LAPTOP-TKKI4B7B:01482] *** An error occurred in MPI_Wait
[LAPTOP-TKKI4B7B:01482] *** reported by process [2196439041,2]
[LAPTOP-TKKI4B7B:01482] *** on communicator MPI_COMM_WORLD
[LAPTOP-TKKI4B7B:01482] *** MPI_ERR_TRUNCATE: message truncated
[LAPTOP-TKKI4B7B:01482] *** MPI_ERRORS_ARE_FATAL (processes in this communicator will now abort,
[LAPTOP-TKKI4B7B:01482] *** and potentially your MPI job)

When the same ctrl and control files are put into a different folder in which I have previously run a separate simulation with an older version of REEF3D and DiveMesh (22.3, if I remember correctly), the simulation runs (albeit the results produced are not correct, the spring does not develop any tension, and the cylinder falls out of the domain).

Any help would be appreciated; I have attached my ctrl and control files along with the SS of the images showing the error and running in the older version.

Hello,rohanjha.
I had the same problem as you during fluid simulation，I am not very clear about the specific cause of this problem，I'm guessing it's due to the Mesh.A similar issue arises when I use different scaling factors for X.Y.Z axis，After I tried to keep the factors as close in proportion as possible, the problem disappeared.Hope my answer can be helpful to you

MPI Error
 

I am also encountering the same error.
The simulation runs with older version 22-03 but doesn't work with new version 23-03.

Can you upload your control files for me to check?

Please find the attached files.


Regards
Keshav Pathak

23.03 had a bug for non-uniform grids. This was fixed in the release candidate of REEF3D. Here is the commit: https://github.com/REEF3D/REEF3D/com...1147dc4e0072be

Just download and compile the RC. Your files run with it.

@Felix: thanks for the info! So problem is already solved.

In general, please always use the latest RC.