Parsim using Cloud Center with an AWS account throughs an error

Hi there,

So I have a Simulink model, which I try to simulate using parsim on a cloudcluster using Mathworks cloudcenter and an AWS account.

I managed to create the cloudcluster, however, when I run the simulation code, it goes through the first teps without problems, but then it sudenly diconnects and terminates the process, here is my script and what I see in the command line:

---------------SCRIPT------------------------

% Simulation Data Path
simModel = 'VSC_steady_state_thermal_overlap';
modelMode = 'consfsw';

%% declare loop variables
last_index = 0;
iteration = 1;


%% Prepare simulation input data
dt_Nm = 10;
dt_Tm = 10^4;
for Nm = -190:dt_Nm:190

    %% calculate the new minimum lifetime
    Initialize;

    for Tm = 0:dt_Tm:250*10^3

        fprintf('\n');
        fprintf('Nm = %d\n', Nm);
        fprintf('Tm = %d\n', Tm);

        if Nm == 0
            para.setpoints.Nm = max(Nm, 0.00001);
        else
            para.setpoints.Nm = Nm;
        end
        para.setpoints.Tm = Tm;

        last_index = last_index + 1;
        InputParameters(last_index) = para;

    end

end


for k = 1:length(InputParameters(1,:))
    in(k) = Simulink.SimulationInput(simModel);
    in(k) = in(k).setModelParameter('SimulationMode', 'accelerator');
    in(k) = in(k).setVariable('para', InputParameters(:,k));
end

% Open a new parallel pool with a specific number of workers, say, 4
c=parcluster; 
setenv('MDCE_DEBUG','true')
pctconfig('preservejobs',true); 
parpool(10);
pctRunOnAll('Initialize');
out = parsim(in,'ShowSimulationManager','on', 'ShowProgress', 'on');

fprintf('\n')
fprintf('Saving simulation data\n')
save(['Steady_State_Thermal_SimData_VSC_' modelMode], 'out', 'InputParameters', '-v7.3')

---------------SCRIPT------------------------

---------------CLI------------------------

Starting parallel pool (parpool) using the 'm6i_12xlarge_48vCPU_192GiB' profile ...

Connected to parallel pool with 10 workers.

Analyzing and transferring files to the workers ...done.

Analyzing and transferring files to the workers ...done.

Analyzing and transferring files to the workers ...done.

Analyzing and transferring files to the workers ...done.

Analyzing and transferring files to the workers ...done.

Analyzing and transferring files to the workers ...done.

Analyzing and transferring files to the workers ...done.

Analyzing and transferring files to the workers ...done.

Analyzing and transferring files to the workers ...done.

Analyzing and transferring files to the workers ...done.

Loaded PLECS module Version 4.7.5

Warning: This version of PLECS is not officially supported for use with MATLAB Version 23.2. Please refer to the manual for a list of the supported platforms.

[26-Feb-2024 08:39:19] Checking for availability of parallel pool...

[26-Feb-2024 08:39:20] Starting Simulink on parallel workers...

[26-Feb-2024 08:41:25] Identifying files required by the model...

[26-Feb-2024 08:41:37] Sending files required by the model to parallel hosts...

[26-Feb-2024 08:44:37] Cleaning up parallel workers...

Error using Run_sim_thermal_steady_state_parsim

The parallel pool has shut down. Use parpool to start a new pool.

Parallel pool using the 'm6i_12xlarge_48vCPU_192GiB' profile is shutting down.

This parallel pool has been shut down.

Caused by:

The parallel pool shut down because the client lost connection to worker 7. Check the network connection or restart the

parallel pool with 'parpool'.

>> c.JobStorageLocation

ans =

'Database on [e-mail]'

---------------CLI------------------------

Notes:

• If I run the same code on local parallel pole, it works just fine
• My Simulink model contains in a addition to standard Simulink blocks a PLECS blockset component, which is a third party tool.
• I tried different AWS instances with different number of vCPUs and RAMs, with no success
• I tried running the code with different number of worker, threads, with no success

Any help is appreciated.

[Edric Ellis]

I suggest you contact MathWorks support for help with this. They can get extra diagnostic information to try and work out what's going wrong.