StreamFlow

The StreamFlow framework is a container-native Workflow Management System (WMS) written in Python 3. It has been designed around two main principles:

• Allow the execution of tasks in multi-container environments, in order to support concurrent execution of multiple communicating tasks in a multi-agent ecosystem.
• Relax the requirement of a single shared data space, in order to allow for hybrid workflow executions on top of multi-cloud or hybrid cloud/HPC infrastructures.

Use StreamFlow

PyPI

The StreamFlow module is available on PyPI, so you can install it using pip.

pip install streamflow

Please note that StreamFlow requires python >= 3.7. Then you can execute it directly from the CLI

streamflow /path/to/streamflow.yml

Docker

StreamFlow Docker images are available on Docker Hub. In order to run a workflow inside the StreaFlow image

• A StreamFlow project, containing a streamflow.yml file and all the other relevant dependencies (e.g. a CWL description of the workflow steps and a Helm description of the execution environment) need to be mounted as a volume inside the container, for example in the /streamflow/project folder
• Workflow outputs, if any, will be stored in the /streamflow/results folder. Therefore, it is necessary to mount such location as a volume in order to persist the results
• StreamFlow will save all its temporary files inside the /tmp/streamflow location. For debugging purposes, or in order to improve I/O performances in case of huge files, it could be useful to mount also such location as a volume
• The path of the streamflow.yml file inside the container (e.g. /streamflow/project/streamflow.yml) must be passed as an argument to the Docker container

The script below gives an example of StreamFlow execution in a Docker container

docker run -d \
    --mount type=bind,source="$(pwd)"/my-project,target=/streamflow/project \
    --mount type=bind,source="$(pwd)"/results,target=/streamflow/results \
    --mount type=bind,source="$(pwd)"/tmp,target=/tmp/streamflow \
    alphaunito/streamflow \
    /streamflow/project/streamflow.yml

Kubernetes

It is also possible to execute the StreamFlow container as a Job in Kubernetes. In this case, StreamFlow is able to deploy Helm models directly on the parent cluster through the ServiceAccount credentials. In order to do that, the inCluster option must be set to true for each involved module on the streamflow.yml file

models:
  helm-model:
    type: helm
    config:
      inCluster: true
      ...

A Helm template of a StreamFlow Job can be found in the helm/chart folder.

Please note that, in case RBAC is active on the Kubernetes cluster, a proper RoleBinding must be attached to the ServiceAccount object, in order to give StreamFlow the permissions to manage deployments of pods and executions of tasks.

Contribute to StreamFlow

StreamFlow uses pipenv to guarantee deterministic builds. Therefore, the recommended way to manage dependencies is by means of the pipenv command.

As a first step, get StreamFlow from GitHub

git clone git@github.com:alpha-unito/streamflow.git

Then you can install all the requred packages using the pipenv command

pip install --user pipenv
cd streamflow
pipenv install

Finally, you can run StreamFlow in the generated virtual environment. In order for this to work, it is necessary to add the streamflow project folder (the one generated by the git clone command) to your PYTHONPATH list

pipenv run python -m streamflow

StreamFlow relies on Travis CI for PyPI and Docker Hub distributions. Therefore, in order to publish a new version of the software, you only have to augment the version number in version.py file.

StreamFlow Team

Iacopo Colonnelli iacopo.colonnelli@unito.it (creator and maintainer)
Barbara Cantalupo barbara.cantalupo@unito.it (maintainer)
Marco Aldinucci aldinuc@di.unito.it (maintainer)

Gaetano Saitta gaetano.saitta@edu.unito.it (contributor)
Alberto Mulone alberto.mulone@edu.unito.it (contributor)