With the code in this repository, you can build a Docker container that provides
the OpenMPI runtime and tools along with various supporting libaries,
including the MPI4Py Python bindings. The container also runs an OpenSSH server
so that multiple containers can be linked together and used via mpirun.
While containers can in principle be started manually via docker run, we suggest that your use
Docker Compose, a simple command-line tool
to define and run multi-container applications. We provde a sample docker-compose.yml
file in the repository:
mpi_head:
image: openmpi
ports:
- "22"
links:
- mpi_node
mpi_node:
image: openmpi
The file defines an mpi_head and an mpi_node. Both containers run the same openmpi image.
The only difference is, that the mpi_head container exposes its SHH server to
the host system, so you can log into it to start your MPI applications.
The following command will start one mpi_head container and three mpi_node containers:
$> docker-compose up -d
$> docker-compose scale mpi_node=16 mpi_head=1
And to /etc/hosts file compiled from docker-compose automagically
$> docker-compose stop mpi_head
$> yes | docker-compose rm -v mpi_head
$> docker-compose up -d
Once all containers are running, connect to mpi_head with:
$> chmod 400 ssh/id_rsa.mpi
$> ssh -i ssh/id_rsa.mpi -p $( source echo_head_port.sh ) tutorial@192.168.99.100
For testing an mpi4py example using the mpi_nodes:
cd mpi4py_benchmarks
cat /etc/hosts | grep mpi_node --color=none | awk '{print $1}' | sort -u > machines && cat ./machines
mpiexec -hostfile machines -n 16 python helloworld.py
For testing dispel4py with mpi mapping:
mpiexec -n 6 -hostfile machines dispel4py mpi dispel4py.examples.graph_testing.pipeline_test