Web interface for Mercury DPM. Allows to run the ProtectiveWall Advanced Tutorial from a browser, running the simulation remotely and displaying the results locally using vtk.js.

Installing

There are quite a few steps:

• Download the repository:

git clone https://github.com/Franzzzzzzzz/MercuryWeb.git
cd MercuryWeb

Installing the back-end server (Mercury DPM + server)

• Download Mercury, saving the sources in a subfolder Mercury_src (for the patches to be applied correctly).

svn checkout https://svn.MercuryDPM.org/SourceCode/Trunk --username guest --password '' Mercury_src

• A few adjustments are required for Mercury to write vtk file in a format understood by vtk.js. Also, the server requires the pthread library to compile, so let's apply a few patches to a few files using the supplied bash script:

cd Mercury_modif
./applypatches.sh
cd ..

• Now we can compile the simulation. To speed things up we can just focus on the ProtectiveWall compilation:

mkdir Mercury_build
cmake ../Mercury_src 
cd Drivers/Tutorials/Advanced/
make 
cd ../../../..
cp Mercury_build/Drivers/Tutorials/Advanced/ProtectiveWall_Server Software

• We also need a couple of files to build the server version of the ProtectiveWall:

cp CppServer/ProtectiveWall_Server.cpp CppServer/Server.h Mercury_src/Drivers/Tutorials/Advanced/

• We do need to also compile the standalone server:

cd CppServer 
g++ -std=c++17 -o ProtectiveWall_Server ProtectiveWall_Server.cpp -lpthread -lstdc++fs 
cp ProtectiveWall_Server ../Software
cd ..

• the Software directory should now contain two executable: ProtectiveWall_Server, receiving and handling the simulation requests from the frontend; and ProtectiveWall, running the actual simulations.

• for safety and to allow for concurrent user, a userlist file is needed, which will allow only identified user to run simulations. 3 basic level of authority are defined, with different allowed max running time (level 0: 1day, level 1: 1hour, level 2: 0s). These can be changed in the ProtectiveWall_Server.cpp file. Let's create a basic userlist file

echo -e 'admin 0\nstudent 1' > userlist.txt

Installing the front-end server (webkit server)

• A 'few' node modules are required. Hopefully this will install them directly:

sudo apt install npm
npm install .

Running

Testing

We can start both server locally and check if everything is ok. We need to run the backend server from a specific folder though:

npm start &
cd dist
../Software/ProtectiveWall_Server &
cd ../..

Now navigating to http://localhost:8080/ will connect to the front-end sever, allow to change the simulation parameters, run the simulation, and display the results. When clicking on the run button, the back-end server is contacted (it can also be contacted directly from the browser at http://localhost:54321/run?Np=50&t=...). If everything looks good, we can deploy:

Deploy

First, we need to adapt a variable in the javascript to fit the back-end server to be contacted for the calculation. Modify the server variable in the file src/index.js accordingly to the IP and port to contact the backend. Default port is 54321, as shown above. The front-end open port is 54322 when using the startserver script as shown below. To start everything:

npm run startserver && cd dist && ../../Software/ProtectiveWall_Server &&

but if everything is working well, we probably also want to be able to disconnect, so running all that with nohup is probably better. But really, a proper server deamon that would restart this programs on crash is the long term / proper solution...