# Workflow template for parsing logs for an experiment group. Can be run as part
# of another workflow, or standalone.

apiVersion: argoproj.io/v1alpha1
kind: WorkflowTemplate
metadata:
  name: log-parsing-workflow
spec:
  serviceAccountName: codex-benchmarks-workflows
  entrypoint: log-parsing-workflow

  # Sadly we need a PVC to share data among steps. This is a limitation of Argo.
  volumeClaimTemplates:
    - metadata:
        name: logs
      spec:
        accessModes: [ "ReadWriteOnce" ]
        resources:
          requests:
            storage: 50Gi
        storageClassName: hcloud-volumes

  arguments:
    parameters: 
      - name: experimentGroupId
      - name: elasticsearchUrl
        value: "https://elastic-es-default-0.elastic-es-default.logging.svc.cluster.local:9200"
      - name: bucket
        value: "codex-benchmarks"
  
  templates:
    - name: log-parsing-workflow
      steps:
        - - name: parse-logs
            template: parse-logs

        - - name: tar-and-upload
            template: tar-and-upload

    - name: parse-logs
      container:
        image: codexstorage/bittorrent-benchmarks:latest
        command: ["poetry", "run", "python", "-m", "benchmarks.cli"]
        args:
          - logs
          - source
          - --output-dir
          - "/var/logs/{{workflow.parameters.experimentGroupId}}"
          - "{{workflow.parameters.experimentGroupId}}"
          - logstash
          - "{{workflow.parameters.elasticsearchUrl}}"
          - --slices
          - "4"

        volumeMounts:
          - name: logs
            mountPath: "/var/logs"

    - name: tar-and-upload
      script:
        image: codexstorage/bittorrent-benchmarks-workflows:latest
        command: ["/bin/bash"]
        source: |
          set -e

          if [ -z "$(ls /var/logs/{{workflow.parameters.experimentGroupId}})" ]; then
            echo "No logs found."
            exit 1
          fi

          echo "Creating tarball."
          tar -czvf \
            "/var/logs/{{workflow.parameters.experimentGroupId}}.tar.gz" \
            -C /var/logs \
            "{{workflow.parameters.experimentGroupId}}"

          echo "Configure s3 alias for endpoint ${AWS_ENDPOINT_URL}."
          mc alias set s3 "${AWS_ENDPOINT_URL}" "${AWS_ACCESS_KEY_ID}" "${AWS_SECRET_ACCESS_KEY}"

          echo "Copy logs."
          mc cp "/var/logs/{{workflow.parameters.experimentGroupId}}.tar.gz"\
            "s3/{{workflow.parameters.bucket}}/logs/{{workflow.parameters.experimentGroupId}}.tar.gz"

        envFrom:
          - secretRef:
              name: s3-codex-benchmarks

        volumeMounts:
          - name: logs
            mountPath: "/var/logs"