Skip to content

Latest commit

 

History

History
45 lines (31 loc) · 3.12 KB

bwUniCluster20.md

File metadata and controls

45 lines (31 loc) · 3.12 KB

1 Run a UM-Bridge model on the bwUniCluster2.0

In this course we are using the bwUniCluster2.0. To access this HPC follow the instruction for the registration.

To set up UM-Bridge on the HPC, log in to the HPC and follow the instructions from the UM-Bridge Documentation on HPCs.

To run a model on the bwUniCluster2.0 it must be put into a container. You can follow the instructions below to run your model on the HPC UM-Bridge setup.

  1. UM-Bridge server as a Python script: First you have to have to transfere your UM-Bridge server into a Python script (.py). In the script you can loos the first two lines of code:

    import nest_asyncio
nest_asyncio.apply()

    those are only necessary in notebooks.

    On the botom line of your server you need to specify the port as follows: umbridge.serve_models([model], int(os.environ.get("PORT", 4242))).

  2. Install Docker: Go to Docker.com and install Docker on your local machine.

  3. Sign up to Docker Hub: Go to Docker Hub and create an account.

  4. Write a Dockerfile: To create a Docker container you need to write a Dockerfile and place it into the folder next to your Python script.

  5. Build the Docker image: docker build -t <container_name> .

  6. Run the Docker container: To finally run your UM-Bridge server from the Docker container do: docker run -it -p 4242:4242 <container_name>

  7. Upload the container to Docker Hub

    7.1 Tag the Docker image: docker tag <container_name> <yourusername>/<container_name>:latest

    7.2 Log in to Docker Hub: docker login

    7.3 Push the image to Docker Hub: docker push <yourusername>/<container_name>:latest

    7.4 Verify the upload: docker run -it -p 4242:4242 <yourusername>/<container_name>

  8. Upload the container to the HPC

    8.1 Log in to bwUniCluster2.0: ssh <username>@bwunicluster.scc.kit.edu

    8.2 Import the container image with singularity: singularity build <container_name>.sif docker://<yourusername>/<container_name>

    8.3 Start the container with singularity: singularity run --writable-tmpfs --pwd / <container_name>.sif In this command, --pwd / sets the working directory inside the container. If your Dockerfile specifies a different working directory, you must replace / with that path to match the configured directory.

  9. Edit job.sh: Go to the job.sh file in your UM-Bridge HPC setup and add the following line: singularity run --writable-tmpfs --pwd / <container_name>.sif Again make sure that you set the correct working directory.

  10. Run the load-balancer: ./load-balancer

  11. Create a tunnel from your local machine to the HPC: ssh <username>@bwunicluster.scc.kit.edu -N -f -L 4242:localhost:4242

  12. Connect your UM-Bridge client through the tunnel by simply running the client on your local machine on port 4242.