[sinhavivekananda318]

I am facing a strange error.

I am running my case with 80 processors.

Now when I am working with particles and a rotating patch called AMI (arbitary meshing interface), Openfoam wants the AMI to be in a single processor.

I did that with toposet utility.

Now, when I am trying to run it with parallel with scotch decomposition method , I am getting the following error.

Fatal error in MPI_Recv: Message truncated, error stack:
MPI_Recv(224).......................: MPI_Recv(buf=0x7fff962cc038, count=8, MPI_BYTE, src=26, tag=1, MPI_COMM_WORLD, sta$
MPIDI_CH3U_Request_unpack_uebuf(618): Message truncated; 24 bytes received but buffer size is 8

Strangely, when I am reducing the processors to 40 and changed the decomposition method to simple, it worked.

But again, with 40 proc and scotch or hierarchical method, it throws the same error.

In summary,

Processor Decomposition Error
80 scotch yes
80 simple yes
40 scotch/hierarchical yes
40 simple NO

I think the issue might be with MPI.

I checked the following link


Message truncated, error stack: MPIDI_CH3U_Receive_data_found

Here, the issue was solved by changing the comms to nonBlocking in the /apps/material/Openfoam/el7/../.../etc/controDict.

I also checked it and found that it's saved as nonBlocking.

Can anyone help?