mpirun hangs when networkingMode=mirrored

[dhtzs]

Windows Version

Microsoft Windows [Version 10.0.22631.3155]

WSL Version

2.1.3.0

Are you using WSL 1 or WSL 2?

• WSL 2
• WSL 1

Kernel Version

5.15.146.1-2

Distro Version

Ubuntu 23.04

Other Software

mpirun (Open MPI) 4.1.4

Repro Steps

1. Create .wslconfig file with the following content:

[wsl2]
dnsTunneling=true
networkingMode=mirrored
firewall=true
autoProxy=true

[experimental]
autoMemoryReclaim=dropcache

2. Terminate WSL using wsl.exe --shutdown and start a new WSL instance.

3. Compile the following C program using mpicc mpi.c -o mpi:

#include <stdio.h>
#include <stdlib.h>
#include "mpi.h"

#define ARRAY_SIZE 16

int main(int argc, char *argv[]) {
    int rank, size;
    int A[ARRAY_SIZE] = {2,5,6,1,6,7,3,9,5,0,6,8,3,5,6,1};
    int chunk_size, remainder, local_chunk_size, local_start, local_sum, total_sum;

    MPI_Init(&argc, &argv);
    MPI_Comm_rank(MPI_COMM_WORLD, &rank);
    MPI_Comm_size(MPI_COMM_WORLD, &size);

    chunk_size = ARRAY_SIZE / size;
    remainder = ARRAY_SIZE % size;

    if (rank < remainder) {
        local_chunk_size = chunk_size + 1;
        local_start = rank * local_chunk_size;
    } else {
        local_chunk_size = chunk_size;
        local_start = rank * chunk_size + remainder;
    }

    local_sum = 0;
    for (int i = local_start; i < local_start + local_chunk_size; i++) {
        local_sum += A[i];
    }

    MPI_Reduce(&local_sum, &total_sum, 1, MPI_INT, MPI_SUM, 0, MPI_COMM_WORLD);

    if (rank == 0) {
        printf("Total sum: %d\n", total_sum);
    }

    MPI_Finalize();
    return 0;
}

4. Execute the program using mpirun -np 2 ./mpi and notice that mpirun hangs.

Expected Behavior

Actual Behavior

Diagnostic Logs

No response

[sadaisystems]

Same problem here

[anak1st]

Same problem

[chanpreetdhanjal]

Hi. Can you please collect networking logs by following the instructions below?
https://github.com/microsoft/WSL/blob/master/CONTRIBUTING.md#collect-wsl-logs-for-networking-issues

[dhtzs]

Hi @chanpreetdhanjal, thanks for the heads up. I reproduced the problem and collected the networking logs following the provided instructions. I hope they can help resolve the issue.

WslNetworkingLogs-2024-06-19_12-15-44.zip

[github-actions]

Diagnostic information

.wslconfig found
Detected appx version: 2.2.4.0
optional-components.txt not found

[CatalinFetoiu]

hello @dhtzs . thanks for reporting the issue and sending the logs! I was able to repro the issue

this is the same symptom as #10855. MPI appears to connect to 127.0.0.1:6001, because of the RST issue mentioned in the other issue, the connection hangs

I will resolve it as a duplicate of that issue

[CatalinFetoiu]

/dupe 10855

[wang1zhen]

Though it might be a dupe, it's still a existing problem and I feel maybe it would be better to leave it open?

[BoqianBIT]

Same problem