A page on containerizing LMRt

Now that we have containerized LMRt, we can start communicating with it. For this step we will want to leave the current directory we’re in and make a new folder named client. This can be anywhere.

1. Prepare files for the container

For LMRt we are only going to send one file which is our configs.yml file. Sending it into the container, rather than building the container with the file present, will allow us to make changes to our configuration without having to completely rebuild the container.

You can directly download the file from here

Your directory should now look like below

emily@VM:~/.../client$ ls
configs.yml

2. Create our metadata file

In order for our controller to communicate properly with the container, you need to create a metadata file. This is in JSON format and gives the model your parameters as well as input files. Below is an example of our metadata.json file:

  1. Our first parameter is recon_iterations which determines how many times we iterate through an algorithm the model depends on.

  2. Our second parameter is figure_type; here we can specify if we want a graph or map to be produced.

  3. Our final parameter is the directory where we want our output files to be saved in ./recon.

For the inputs, we are only sending our config.yml file found here

{
    "parameters": {
        "recon_iterations": 1,
        "figure_type": "graph",
        "job_dirpath": "./recon"
    },
    "inputs": {
        "configs": "configs.yml"
    }
}

Your directory should now look like

emily@VM:~/.../client$ ls
configs.yml  metadata.json

3. Run the model

We are now ready to run our container using our presto run <image name> command. After the model finishes you should receive a zip file that has the output files from the model.

If you experience errors at this step and need to do presto run again, make sure you are doing Step 5 at the same time

In a separate terminal, you can run docker logs --follow $(docker ps -q) after you run the below command to follow what is happening inside of the container in real time.

presto run lmrt

emily@VM:~/.../client$ presto run demo
Running the container...
{'Status': 'running', 'Running': True, 'Paused': False, 'Restarting': False, 'OOMKilled': False, 'Dead': False,
'Pid': 86211, 'ExitCode': 0, 'Error': '', 'StartedAt': '2022-03-11T00:58:10.323020527Z', 'FinishedAt': '0001-01-01T00:00:00Z'}
Output files successfully saved at ./response_data.zip

4. View reconstruction files

When you check your files you should see a new zip archive response_data.zip. You can then unzip the archive and view your output files. This should be an image of a figure and the recon folder.

5. Clean up the container

It is important that you do not skip this step. Click next to access the instructions.

Next

Tags: