Skip to content

doubtfire-lms/doubtfire-overseer

Repository files navigation

Deployment steps

  • Clone the repo.

  • Create .env file with the following variables and required values:

    RABBITMQ_HOSTNAME=192.254.254.254
    RABBITMQ_USERNAME=guest
    RABBITMQ_PASSWORD=guest
    
    HOST_XFS_VOLUME=/some/path
    CONTAINER_NAME=Some-Docker-Container-Name
    • RABBITMQ_HOSTNAME, RABBITMQ_USERNAME, and RABBITMQ_PASSWORD provide details needed to access the rabbitmq.
    • HOST_XFS_VOLUME is a path on the host that will run the automated code. This needs to refer to the location where overseer containers will store their work. Overseer uses this to setup a volume in each container it creates. When multiple overseer instances are run this path needs to be unique for each instance. Overseer will create sandbox and output folders in this location.
    • CONTAINER_NAME is optional (defaults to overseer-container) and provides the name of the container that overseer creates. If there are multiple overseer instances then they must have unique container names.
  • Easiest option is to setup rabbitmq alongside overseer in the same docker network, and have it setup to run with the management interface.

  • Alternatively:

    • Configure your machine to add an alias IP address to en0 (or whatever network device you are using) as the value specified to the key RABBITMQ_HOSTNAME in the .env file before. On MacOS, you can use: sudo ifconfig en0 alias 192.254.254.254 255.255.255.0

    • Run RabbitMQ server:

      docker run -d --hostname my-rabbit --name some-rabbit -p 5672:5672 -p 15672:15672 rabbitmq:3

    • Enable RabbitMQ management plugin:

      docker exec -t some-rabbit rabbitmq-plugins enable rabbitmq_management

  • Run overseer using: bundle exec ruby ./app.rb

Basic operation

  • When overseer starts is listens to the q.tasks queue on the ontrack exchange of the rabbitmq and awaits any messages marked as task.submission. When a submission is received the message triggers action in overseer.
  • Params received include:
    • output_path: where to store automated assessment results
    • submission: the path to the submission zip file or folder
    • assessment: the path to the submission assessment resources
    • timestamp: the time of the submission
    • task_id: the task associated with the submission
    • overseer_assessment_id: the overseer assessment id
    • docker_image_name_tag: the tag of the docker image to run the scripts within
  • Overseer starts by unpacking the submission along with the assessment resources to the working directory. Files in the assessment resources will override those in the submission if needed.
  • Once the files are in place Overseer starts a container with the indicated CONTAINER_NAME using the image details from the assessment resources and runs two scripts: build.sh is run and if the result is 0 it runs run.sh.
    • Anything written to standard output and standard error from these scripts will be collected in the output message for the task.
    • Each of these scripts is passed a path to a yaml file where comment and status changes are stored.
    • build.sh should write messages to the build_message key in the yaml file.
    • run.sh should write messages to the run_message key in the yaml file.
    • Both build and run scripts can write a new status to new_status.
    • The exit status of the build and run steps is appended to the output text collected from standard out.
  • When both scripts have been run, a message is returned to Doubtfire/OnTrack using the overseer.result routing key to the q.overseer of the ontrack exchange. This contains the task id, timestamp, and overseer assessment id. The output text and yaml files are also available to OnTrack.