controller: emit 5-sketch routing-connector pipeline (DDSketch/KLL/HLL/CountSketch/CountMinSketch)

[zzylol]

Summary

Rewrites the typed L5 edge YAML emitter (controller/src/config/stage_config.rs::emit_edge_yaml) so the agent loads all 5 sketch processors and the OTel routing connector (NOT the deprecated routing processor) dispatches per metric to the right sketch family — the canonical wire shape for issue #46.

Bugfix call-out: routing is a CONNECTOR, not a processor

OTel collector v0.106+ removed routingprocessor and re-shipped the routing component as a connector under a top-level connectors: block. The legacy emit path placed routing under processors:, which fails confmap validation at agent boot:

error decoding 'processors': unknown type: "routing"

The asap-otel binary's builder-config.yaml registers routingconnector. This PR emits the connector-form layout that build expects.

Canonical YAML shape

receivers:
  otlp: ...
processors:
  ddsketchprocessor: ...
  kllprocessor: ...
  hllprocessor: ...
  countsketchprocessor: ...
  countminsketchprocessor: ...
  gorillas3: ...     # when archive_tier_metrics non-empty
  batch: ...
connectors:
  routing:
    default_pipelines: [metrics/raw_passthrough]
    table:
      - statement: 'route() where metric.name == "http_latency_ms"'
        pipelines: [metrics/ddsketch_path]
      - statement: 'route() where metric.name == "request_size_bytes"'
        pipelines: [metrics/kll_path]
      - statement: 'route() where metric.name == "unique_users_per_min"'
        pipelines: [metrics/hll_path]
      - statement: 'route() where metric.name == "top_endpoint_qps"'
        pipelines: [metrics/countsketch_path]
      - statement: 'route() where metric.name == "endpoint_request_freq"'
        pipelines: [metrics/countminsketch_path]
exporters:
  otlp/backend: ...
service:
  pipelines:
    metrics:                              # entry
      receivers: [otlp]
      exporters: [routing]
    metrics/raw_passthrough:              # default
      receivers: [routing]
      processors: [gorillas3?, batch]
      exporters: [otlp/backend]
    metrics/ddsketch_path:
      receivers: [routing]
      processors: [gorillas3?, ddsketchprocessor, batch]
      exporters: [otlp/backend]
    metrics/kll_path: ...                 # same shape
    metrics/hll_path: ...
    metrics/countsketch_path: ...
    metrics/countminsketch_path: ...

The 6 named pipelines

1. metrics — entry (receiver otlp, exporter routing)
2. metrics/raw_passthrough — default fall-through (e.g. http_requests_total, freshness probes)
3. metrics/ddsketch_path
4. metrics/kll_path
5. metrics/hll_path
6. metrics/countsketch_path
7. metrics/countminsketch_path

(Plus metrics/prometheus_archive when Mode-3 metrics are also configured.)

EdgeStageConfig field

Adds metric_to_family: HashMap<String, SketchKind> to EdgeStageConfig (agreed convention with the parallel planner agent). Empty map ⇒ legacy single-pipeline / Mode-3 / warm-passthrough emit shapes preserved verbatim.

gorillas3 invariant

gorillas3 runs FIRST in every per-sketch pipeline so the cold-tier write happens BEFORE the family processor mutates / suffix-renames the stream — same invariant the legacy emit path enforces, pinned by mvp46_per_sketch_pipelines_have_gorillas3_first_when_archive_declared.

Freshness-probe routing (PR #333)

warm_passthrough_metrics route to metrics/raw_passthrough so the metric name is preserved end-to-end (no DDSketch _quantile suffix). Pinned by mvp46_warm_passthrough_routes_to_raw_passthrough_pipeline.

Test plan

• All 99 config::* tests pass
• 11 new MVP §46 tests pin the new wire shape:
  • mvp46_emit_loads_all_5_sketch_processors
  • mvp46_routing_lives_in_connectors_not_processors (the bugfix)
  • mvp46_emits_all_6_named_pipelines
  • mvp46_entry_pipeline_routes_to_connector_not_processor
  • mvp46_per_sketch_pipelines_use_routing_as_receiver
  • mvp46_per_sketch_pipelines_have_gorillas3_first_when_archive_declared
  • mvp46_routing_table_dispatches_per_metric_to_correct_family
  • mvp46_default_pipeline_is_raw_passthrough
  • mvp46_warm_passthrough_routes_to_raw_passthrough_pipeline
  • mvp46_empty_metric_to_family_falls_back_to_legacy_emit
  • mvp46_composes_with_prometheus_archive_mode3
• Pre-existing test failures on main (analyzer / intent_algebra / planner cost_model / opamp / api_tests — 10 total) are untouched

Refs: #46

🤖 Generated with Claude Code