Skip to content

Latest commit

 

History

History
45 lines (34 loc) · 1.43 KB

README.md

File metadata and controls

45 lines (34 loc) · 1.43 KB
Raw

Running Workflows via GMS

This codebase is primarily concerned with first-time setup and providing required resources to operate the GMS integration. Details on how to use these resources through the GMS interface are provided on the GMS wiki

Initial Setup

Create requisite resources. Any that already exist will spit out an "already exists" stderr but these can be safely ignored. This creates the required service accounts, bucket, and permissions.

bash resources.sh --project PROJECT --bucket BUCKET

Alternatively, these files are available via the docker image mgibio/cloudize-workflow:latest in the directory /opt/gms/, so the above command would be

bash /opt/gms/resources.sh --project PROJECT --bucket BUCKET

Add the values given from this script to your environment configuration.

You should now be able to run a workflow on the cloud via your new environment configuration.

Who can use these resources?

Permissions required to run a GMS workflow are:

  • create a compute instance
  • read/write files to bucket
  • use Cromwell server service account which enables using the Cromwell compute service account, and run lifesciences workflows

GMS should be enforcing these requirements on the running user, meaning you have to be gcloud authd to operate. This works for manual workflow kickoffs but not for automated ingestion, which will be tackled later.