calrissian 0.5.0

CWL runner for Kubernetes

calrissian

CWL on Kubernetes

Overview

Calrissian is a CWL implementation designed to run individual steps as Pods in a kubernetes cluster

It is in development and includes a simple workflow (revsort single / array) to reverse and sort the contents of text files.

Preparing Openshift

The openshift/ directory contains YAML files that demonstrate the basic functionality. To configure Openshift, do the following:

1. Create a project in openshift (e.g. calrissian-demo-project). Note that namespaces must be unique, so if calrissian-demo-project is already in use on the cluster, you will need to choose a different name.

    oc new-project calrissian-demo-project
   

2. Create a pair of roles in your project to allow managing kubernetes pods and logs:

    oc create role pod-manager-role --verb=create,delete,list,watch --resource=pods
    oc create role log-reader-role --verb=get,list --resource=pods/log
   

3. Bind your project's default service account access to these roles. Calrissian running inside the cluster will use this service account to make Kubernetes API calls.

    oc create rolebinding pod-manager-default-binding --role=pod-manager-role --serviceaccount=calrissian-demo-project:default
    oc create rolebinding log-reader-default-binding --role=log-reader-role --serviceaccount=calrissian-demo-project:default
   

4. Create the BuildConfig - this allows Openshift to build the source code into a Docker image and starts a build automatically. If you wish to build a different branch or repo, edit this file.

    oc create -f openshift/BuildConfig.yaml
   

5. Create the VolumeClaims - these are the storage locations that will be shared between pods, and must support read-write many access.

    oc create -f openshift/VolumeClaims.yaml
   

Running the example

With the VolumeClaims and Build processes in place, you can run the example with 2 jobs. These jobs use the docker image built by the BuildConfig, so they cannot be run until the build completes.

1. Stage the input data and workflow onto the input-data volume with the StageInputDataJob.

    oc create -f openshift/StageInputDataJob.yaml
   

2. Run the workflow engine with the CalrissianJob

    oc create -f openshift/CalrissianJob-revsort.yaml
   

3. Job pods can be monitored from the openshift web interface or via command-line

    oc logs -f job/calrissian-revsort-array
   

Notes

1. Calrissian will delete completed jobs for individual steps, but you may want to delete the calrissian-revsort-array job manually after running it.