Deploying Cloud-Based Automated Continuous Integration to Assist in the Community Development of GFDL's Finite Volume Cubed-Sphere Dynamical Core
A major challenge to community code development and management is the testing and validation of public contributions. The community-developed GFDL Finite-Volume Cubed Sphere Dynamical Core (FV3) is no exception: automated testing of contributions made to the FV3 public repository is paramount for ensuring the integrity of the many earth-system models and forecasting applications using FV3 as a dynamical core. A build and test system for the FV3 dynamical core was developed for internal testing on NOAA Research and Development High Performance Computing Systems (RDHPCS). We have designed a continuous integration (CI) approach for the FV3 dynamical core Github repository that uses a cloud-based platform to perform automated compilation and reproducibility testing to validate community code contributions. A combination of NOAA RDHPCS Parallel Works virtual machines and containers developed at GFDL are used to compile and test code on the cloud efficiently. We will also discuss how we adapted the FV3 tests for automated CI.