feat(pipelines): migrate analytics to the pipelines aggregation engine

What does this MR do and why?

Migrates pipeline analytics from the finishedPipelines aggregation engine to the new pipelines engine (gitlab!241321 (merged)), backed by siphon_p_ci_pipelines instead of the deprecated ci_finished_pipelines table. The new table holds all pipeline states, not just finished ones.

The GLQL query syntax is unchanged — existing queries keep compiling and running — but result semantics change: the engine now covers all pipeline states, so totalCount (and any query not filtered to terminal statuses) now includes running/pending pipelines. The rate metrics keep their old meaning: their denominator remains completed pipelines. Mechanically:

• The four rate metrics keep their names and compile to aliased selections of the engine's parameterised outcomeRate metric: successRate: outcomeRate(status: "success"). Response keys therefore still match the GLQL field names, so the transformer and the fields metadata contract need no changes.
• Sorting by a rate emits orderBy: [{identifier: "outcomeRate", parameters: {status: "..."}}] via a new graphql_sort_params analyzer hook (default empty), consulted only when the sort field has no user-resolved parameters to inherit. Verified live that the backend disambiguates duplicate outcomeRate order identifiers by their parameters.
• The status filter widens from the four terminal statuses to the full CI status set (via the shared ci_status_field_type()), reverting the !405 (merged) restriction — status = "running" analytics now work.

Docs update: gitlab!244089 (merged)

Agent update: gitlab-org/modelops/applied-ml/code-suggestions/ai-assist!6116 (merged)

Example Usage

Build the Ruby extension once:

cd glql_rb
bundle install
bundle exec rake compile

Rate metrics with a rate sort (aliased outcomeRate selections)

mode: analytics
query: type = Pipeline and project = "gitlab-org/gitlab" and started > 2026-06-01
dimensions: ref
metrics: totalCount, successRate, failureRate, durationQuantile
sort: successRate desc

Test via ruby extension

GLQL_QUERY='type = Pipeline and project = "gitlab-org/gitlab" and started > 2026-06-01'

echo "q = Glql.compile('${GLQL_QUERY}', { mode: 'analytics', dimensions: 'ref', metrics: 'totalCount, successRate, failureRate, durationQuantile', sort: 'successRate desc' })[\"output\"]; puts q; File.write('/tmp/glql_query.json', { query: q, variables: { limit: 3 } }.to_json); exit" | bundle exec rake console && curl -s -X POST https://gitlab.com/api/graphql -H "Content-Type: application/json" -d @/tmp/glql_query.json | python3 -m json.tool

Generated GraphQL:

query GLQL($before: String, $after: String, $limit: Int) {
  project(fullPath: "gitlab-org/gitlab") {
    analytics {
      pipelines(startedAtFrom: "2026-06-01 23:59") {
        aggregated(before: $before, after: $after, first: $limit, orderBy: [{direction: DESC, identifier: "outcomeRate", parameters: {status: "success"}}]) {
          count
          nodes {
            dimensions {
              ref
            }
            totalCount
            successRate: outcomeRate(status: "success")
            failureRate: outcomeRate(status: "failed")
            durationQuantile(quantile: 0.95)
          }
        }
      }
    }
  }
}

Response:

{
    "data": {
        "project": {
            "analytics": {
                "pipelines": {
                    "aggregated": {
                        "count": 26468,
                        "nodes": [
                            {
                                "dimensions": { "ref": "refs/workloads/affdbaf60ae" },
                                "totalCount": 1,
                                "successRate": 1.0,
                                "failureRate": 0.0,
                                "durationQuantile": 246.0
                            },
                            {
                                "dimensions": { "ref": "refs/workloads/429058c68c4" },
                                "totalCount": 1,
                                "successRate": 1.0,
                                "failureRate": 0.0,
                                "durationQuantile": 241.0
                            },
                            {
                                "dimensions": { "ref": "refs/workloads/34e106f4ac7" },
                                "totalCount": 1,
                                "successRate": 1.0,
                                "failureRate": 0.0,
                                "durationQuantile": 288.0
                            }
                        ]
                    }
                }
            }
        }
    }
}

Non-terminal status filter (new — impossible with finishedPipelines)

mode: analytics
query: type = Pipeline and project = "gitlab-org/gitlab" and status = "running"
dimensions: status, started(monthly)
metrics: totalCount
sort: started desc

Test via ruby extension

GLQL_QUERY='type = Pipeline and project = "gitlab-org/gitlab" and status = "running"'

echo "q = Glql.compile('${GLQL_QUERY}', { mode: 'analytics', dimensions: 'status, started(monthly)', metrics: 'totalCount', sort: 'started desc' })[\"output\"]; puts q; File.write('/tmp/glql_query.json', { query: q, variables: { limit: 3 } }.to_json); exit" | bundle exec rake console && curl -s -X POST https://gitlab.com/api/graphql -H "Content-Type: application/json" -d @/tmp/glql_query.json | python3 -m json.tool

Generated GraphQL:

query GLQL($before: String, $after: String, $limit: Int) {
  project(fullPath: "gitlab-org/gitlab") {
    analytics {
      pipelines(status: ["running"]) {
        aggregated(before: $before, after: $after, first: $limit, orderBy: [{direction: DESC, identifier: "startedAt", parameters: {granularity: "monthly"}}]) {
          count
          nodes {
            dimensions {
              status
              startedAt(granularity: "monthly")
            }
            totalCount
          }
        }
      }
    }
  }
}

Response:

{
    "data": {
        "project": {
            "analytics": {
                "pipelines": {
                    "aggregated": {
                        "count": 44,
                        "nodes": [
                            {
                                "dimensions": { "status": "running", "startedAt": "2026-07-01" },
                                "totalCount": 53
                            },
                            {
                                "dimensions": { "status": "running", "startedAt": "2026-06-01" },
                                "totalCount": 6
                            },
                            {
                                "dimensions": { "status": "running", "startedAt": "2026-05-01" },
                                "totalCount": 9
                            }
                        ]
                    }
                }
            }
        }
    }
}

How to set up and validate locally

1. cargo test — full suite, including the new aliased-rendering, rate-sort, and non-terminal status filter cases in tests/pipeline_analytics_tests.rs.
2. DUMP_GRAPHQL=1 cargo test && npm run test:graphql — validates every generated query against the fetched GitLab schema (refresh it first with ./scripts/fetch_gitlab_graphql_schema.sh).
3. Run either Example Usage snippet above against gitlab.com/api/graphql.