Spike: PMG query optimization

While we are building out the GraphQL pipeline mini graph, we need to look into whether our queries still make sense.

We are considering consolidating getPipelineStages and getLinkedPipelines into one query getPipelineMiniGraph for our initial query. Considerations exist in the architectural blueprint.

Note - This is simply the approach for the initial rendering of the component. All other data (read: dropdown data) will occur farther down the chain in the implementation.

Current Queries

Pipeline Stages

query getPipelineStages($fullPath: ID!, $iid: ID!) {
  project(fullPath: $fullPath) {
    id
    pipeline(iid: $iid) {
      id
      stages {
        nodes {
          id
          name
          detailedStatus {
            id
            icon
            group
          }
        }
      }
    }
  }
}

Linked pipelines

query getLinkedPipelines($fullPath: ID!, $iid: ID!) {
  project(fullPath: $fullPath) {
    id
    pipeline(iid: $iid) {
      id
      path
      downstream {
        nodes {
          id
          path
          project {
            id
            name
          }
          detailedStatus {
            id
            group
            icon
            label
          }
          sourceJob {
            id
            retried
          }
        }
      }
      upstream {
        id
        path
        project {
          id
          name
        }
        detailedStatus {
          id
          group
          icon
          label
        }
      }
    }
  }
}

Proposed Query

query getPipelineMiniGraph() {
  project(fullPath: $fullPath) {
    id
    pipeline(iid: $iid) {
      id
      path
      downstream {
        nodes {
          id
          path
          project {
            id
            name
          }
          detailedStatus {
            id
            group
            icon
            label
          }
          sourceJob {
            id
            retried
          }
        }
      }
      upstream {
        id
        path
        project {
          id
          name
        }
        detailedStatus {
          id
          group
          icon
          label
        }
      }
      stages {
        nodes {
          id
          name
          detailedStatus {
            id
            icon
            group
          }
        }
      }
    }
  }
}

Other options include:

• Leaving the queries as-is but making sure they exist in the same directory
• Breaking down the queries even further

Our path forward will be determined by what we learn from experimenting with the query optimization.