Run SOMISANA forecast models #567
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| name: Run SOMISANA forecast models | |
| on: | |
| workflow_dispatch: | |
| inputs: | |
| run_date: | |
| description: Run date (default = today). yyyymmdd | |
| required: false | |
| default: | |
| # commenting the schedule for now until we are ready to deploy this workflow operationally | |
| schedule: | |
| - cron: '0 02 * * *' # 0400 SAST | |
| # Many of the env variables reference the current branch, | |
| # which is set dynamically in the branch-ref job below | |
| # So the env variables are mostly set in the envs job below the branch-ref job | |
| env: | |
| REGISTRY: ghcr.io | |
| jobs: | |
| # Dynamically set the branch ref to the currently executing branch | |
| branch-ref: | |
| runs-on: ubuntu-latest | |
| outputs: | |
| value: ${{ steps.BRANCH_REF.outputs.value }} | |
| steps: | |
| - name: Set the BRANCH_REF | |
| id: BRANCH_REF | |
| run: | | |
| echo "value=${GITHUB_REF##*/}" >> $GITHUB_OUTPUT | |
| # Set the environment variables using the current branch reference | |
| envs: | |
| needs: [branch-ref] | |
| runs-on: ubuntu-latest | |
| env: | |
| RUN_DATE: ${{ inputs.run_date }} | |
| outputs: | |
| BRANCH_REF: ${{ needs.branch-ref.outputs.value }} | |
| TOOLKIT_IMAGE: ${{ steps.ENVS.outputs.TOOLKIT_IMAGE }} | |
| MODEL_RUN_DATE: ${{ steps.ENVS.outputs.MODEL_RUN_DATE }} | |
| EEZ_DATA_DIR: ${{ steps.ENVS.outputs.EEZ_DATA_DIR }} | |
| steps: | |
| - name: Configure run date | |
| id: run_date | |
| run: | | |
| echo "yyyymmdd=${RUN_DATE:=$(date +'%Y%m%d')}" >> $GITHUB_OUTPUT | |
| - name: Set envs | |
| id: ENVS | |
| run: | | |
| echo "TOOLKIT_IMAGE=${{ github.repository }}_toolkit_${{ needs.branch-ref.outputs.value }}" >> $GITHUB_OUTPUT | |
| echo "MODEL_RUN_DATE=${{ steps.run_date.outputs.yyyymmdd }}" >> $GITHUB_OUTPUT | |
| echo "EEZ_DATA_DIR=/home/runner/somisana/global_data/${{ steps.run_date.outputs.yyyymmdd }}/${{ needs.branch-ref.outputs.value }}" >> $GITHUB_OUTPUT # Think it would have been better to have $BRANCH_REF after somisana i.e. somisana/$BRANCH_REF but doing it this way to be consistent with how the model run directories were set up. Maybe we can change it later... | |
| # The SOMISANA toolkit is a suite of scripts to facilitate | |
| # running SOMISANA models - for example the Algoa Bay Forecast. | |
| # In the context of a testing environment, where updates to these | |
| # scripts is part of the development workflow, the toolkit needs | |
| # to be compiled on every run | |
| toolkit: | |
| needs: [envs] | |
| runs-on: ubuntu-latest | |
| env: | |
| BRANCH_REF: ${{ needs.envs.outputs.BRANCH_REF }} | |
| TOOLKIT_IMAGE: ${{ needs.envs.outputs.TOOLKIT_IMAGE }} | |
| SHA: sha-${{ github.sha }} | |
| outputs: | |
| image: ${{ env.REGISTRY }}/${{ steps.lowercase.outputs.image_name }} | |
| steps: | |
| - name: Get image name lowercase | |
| id: 'lowercase' | |
| run: | | |
| IMAGE_NAME_LOWER=$(echo $TOOLKIT_IMAGE | tr '[:upper:]' '[:lower:]') | |
| echo "image_name=$IMAGE_NAME_LOWER" >> $GITHUB_OUTPUT | |
| - name: Check out source code | |
| uses: actions/checkout@main # change to main | |
| with: | |
| ref: ${{ env.BRANCH_REF }} | |
| - name: Log in to the Container registry | |
| uses: docker/login-action@master | |
| with: | |
| registry: ${{ env.REGISTRY }} | |
| username: ${{ github.actor }} | |
| password: ${{ secrets.GITHUB_TOKEN }} | |
| - name: Check if image with sha tag already exists | |
| id: check-image | |
| run: | | |
| set -e | |
| SHA_IMAGE=$(docker images -q ${REGISTRY}/${{ steps.lowercase.outputs.image_name }}:${SHA} || true) | |
| if [ -z "$SHA_IMAGE" ]; then | |
| echo "image_exists=false" >> $GITHUB_OUTPUT | |
| else | |
| echo "image_exists=true" >> $GITHUB_OUTPUT | |
| fi | |
| - name: Extract metadata (tags, labels) for Docker | |
| id: meta | |
| uses: docker/metadata-action@master | |
| with: | |
| images: ${{ env.REGISTRY }}/${{ env.TOOLKIT_IMAGE }} | |
| tags: | | |
| latest | |
| ${{ env.SHA }} | |
| - name: Build and push | |
| uses: docker/build-push-action@master | |
| with: | |
| context: toolkit | |
| push: true | |
| tags: ${{ steps.meta.outputs.tags }} | |
| labels: ${{ steps.meta.outputs.labels }} | |
| # Download environmental data used for model boundaries and surface forcing. | |
| # This job is retried if it fails, since the failure can be due to upstream errors | |
| # These downloads are used to create forcing files that are fed as input to the CROCO model | |
| # => marine.copernicus.eu: This is ocean data that forms the lateral boundaries of our model run | |
| # => ncei.noaa.gov: This is atmospheric data used to create surface boundary conditions for our model run | |
| download-global-data: | |
| needs: [toolkit, envs] | |
| runs-on: somisana | |
| env: | |
| EEZ_DATA_DIR: ${{ needs.envs.outputs.EEZ_DATA_DIR }} | |
| MODEL_RUN_DATE: ${{ needs.envs.outputs.MODEL_RUN_DATE }} | |
| SHA: sha-${{ github.sha }} | |
| BRANCH_REF: ${{ needs.envs.outputs.BRANCH_REF }} | |
| steps: | |
| - name: Clean /home/runner/somisana/global_data/ | |
| run: >- | |
| find \ | |
| /home/runner/somisana/global_data/* \ | |
| -maxdepth 0 \ | |
| -type d \ | |
| -ctime +5 \ | |
| -exec \ | |
| rm \ | |
| -rf {} \; | |
| - name: Configure no. retries | |
| id: retries | |
| run: | | |
| if [ "$BRANCH_REF" = "stable" ] | |
| then | |
| echo "RETRIES=30" >> $GITHUB_OUTPUT | |
| else | |
| echo "RETRIES=1" >> $GITHUB_OUTPUT | |
| fi | |
| - name: Source ~/.bashrc (for NVM - required for non-interactive shells) | |
| continue-on-error: true # Maybe Node.js exists anyway | |
| run: | | |
| source ~/.bashrc | |
| nvm use 19.6.0 | |
| - name: Create ${{ env.EEZ_DATA_DIR }} directory | |
| continue-on-error: true | |
| run: | | |
| # rm -rf ${{ env.EEZ_DATA_DIR }} # not sure why we need this... just commenting for now due to issues with nomads, and our code doesn't pick up when randomly don't get downloaded with status code 404. So just re-running the job manually until I have time to fix it properly. And we need the previously downloaded files, otherwise it fails again | |
| mkdir -p ${{ env.EEZ_DATA_DIR }} | |
| chown -R :runners ${{ env.EEZ_DATA_DIR }} | |
| chmod -R 774 ${{ env.EEZ_DATA_DIR }} | |
| - name: Download GFS | |
| uses: nick-fields/retry@master | |
| with: | |
| timeout_minutes: 30 # Script is considered failed if this limit is reached | |
| retry_wait_seconds: 300 # Wait 5 minutes and try again | |
| max_attempts: ${{ steps.retries.outputs.RETRIES }} | |
| retry_on: any | |
| warning_on_retry: true | |
| shell: bash | |
| continue_on_error: false | |
| on_retry_command: rm -f ${{ env.EEZ_DATA_DIR }}/*grb* | |
| command: >- | |
| docker run \ | |
| --rm \ | |
| -v ${{ env.EEZ_DATA_DIR }}/:/tmp/somisana/current \ | |
| ${{ needs.toolkit.outputs.image }}:${{ env.SHA }} \ | |
| download \ | |
| --provider gfs \ | |
| --workdir /tmp/somisana/current \ | |
| --matlab-env /tmp/somisana/current/.env \ | |
| --download-date ${{ env.MODEL_RUN_DATE }} \ | |
| --domain 11,36,-39,-25 | |
| - name: Download Mercator | |
| uses: nick-fields/retry@master | |
| with: | |
| timeout_minutes: 60 # Script is considered failed if this limit is reached | |
| retry_wait_seconds: 300 # Wait 5 minutes and try again | |
| max_attempts: ${{ steps.retries.outputs.RETRIES }} | |
| retry_on: any | |
| warning_on_retry: true | |
| shell: bash | |
| continue_on_error: false | |
| on_retry_command: rm -f ${{ env.EEZ_DATA_DIR }}/*mercator* | |
| command: >- | |
| docker run \ | |
| --rm \ | |
| -v ${{ env.EEZ_DATA_DIR }}/:/tmp/somisana/current/ \ | |
| -e COPERNICUS_USERNAME=${{ secrets.COPERNICUS_USERNAME }} \ | |
| -e COPERNICUS_PASSWORD=${{ secrets.COPERNICUS_PASSWORD }} \ | |
| ${{ needs.toolkit.outputs.image }}:${{ env.SHA }} \ | |
| download \ | |
| --provider mercator \ | |
| --workdir /tmp/somisana/current \ | |
| --download-date ${{ env.MODEL_RUN_DATE }} \ | |
| --domain 11,36,-39,-25 | |
| # now do the model specific steps | |
| # currently this is done by calling different model reusable workflows in series | |
| # In future we should make a generic model reusable workflow which takes the model name as an input | |
| # Then the model runs can be carried out using a matrix strategy | |
| # (there's an example of this here: https://docs.github.com/en/actions/using-workflows/reusing-workflows#example-reusable-workflow) | |
| call_run_algoa-bay-forecast: | |
| needs: [envs,toolkit,download-global-data] | |
| uses: ./.github/workflows/run_algoa-bay-forecast.yml | |
| with: | |
| BRANCH_REF: ${{ needs.envs.outputs.BRANCH_REF }} | |
| MODEL_RUN_DATE: ${{ needs.envs.outputs.MODEL_RUN_DATE }} | |
| TOOLKIT_IMAGE: ${{ needs.toolkit.outputs.image }}:sha-${{ github.sha }} | |
| EEZ_DATA_DIR: ${{ needs.envs.outputs.EEZ_DATA_DIR }} | |
| secrets: inherit | |
| call_run_false-bay-forecast: | |
| needs: [envs,toolkit,download-global-data] | |
| uses: ./.github/workflows/run_false-bay-forecast.yml | |
| with: | |
| BRANCH_REF: ${{ needs.envs.outputs.BRANCH_REF }} | |
| MODEL_RUN_DATE: ${{ needs.envs.outputs.MODEL_RUN_DATE }} | |
| TOOLKIT_IMAGE: ${{ needs.toolkit.outputs.image }}:sha-${{ github.sha }} | |
| EEZ_DATA_DIR: ${{ needs.envs.outputs.EEZ_DATA_DIR }} | |
| secrets: inherit | |