Execution done twice when application fails

When an OTB application fails, the execution is done twice

Tested with TensorflowModelServe, but this has to be confirmed with other applications

for now I don't understand