master
md 16.4 KB

chartengine

chartengine compiles chart templates and builds deterministic chart plans (create, update, remove) from metrix.Reader snapshots.

Audience: ModuleV2 collector authors and framework contributors.

See also: charttpl (template DSL), metrix (metrics storage and read API).

Purpose

Stage Responsibility
charttpl Template decode/defaults/validation
chartengine.Compile Build immutable program IR
Engine.PreparePlan Prepare plan actions plus explicit commit/abort boundary
chartemit.ApplyPlan Emit plan to Netdata wire protocol

Collector-Facing Contract

For ModuleV2 collectors, the runtime integration expects:

Requirement Why it matters
MetricStore() returns metrix.CollectorStore (cycle-managed) Job runtime controls cycle boundaries and success/failure semantics
ChartTemplateYAML() returns valid charttpl YAML Loaded once at autodetection/post-check
Collector writes metrics during Collect() only Planner runs after a successful cycle commit
Metric names used in template selectors are present in group metric scope Compile/validate consistency

Public API Surface

API Purpose
New(opts...) Create engine with policy/runtime options
Load(spec, revision) / LoadYAML(data, revision) Compile and publish program revision
PreparePlan(reader) Build deterministic action plan from reader snapshot and return an explicit attempt
RuntimeStore() Access chartengine internal runtime metrics store
WithEnginePolicy(...) Configure selector + autogen behavior
WithRuntimeStore(...) Override/disable self-metrics store
WithSeriesSelectionAllVisible() Process all visible series instead of filtering to latest successful collect cycle. Intended for runtime/internal stores that commit immediately (no cycle boundaries).
WithEmitTypeIDBudgetPrefix(...) Set the effective type-id prefix used by autogen budget checks
WithRuntimePlannerMode(...) Enable runtime planner mode with no-write-tick semantics, for jobs/tests that drive planning directly from runtime metrics instead of collect-cycle boundaries.

End-to-End Example (Single Flow)

// 1) Collector writes metrics.
store := metrix.NewCollectorStore()
meter := store.Write().SnapshotMeter("app")
meter.Counter("requests_total").ObserveTotal(100)

// 2) Engine loads chart template.
engine, err := chartengine.New(
    chartengine.WithEnginePolicy(chartengine.EnginePolicy{
        Autogen: &chartengine.AutogenPolicy{Enabled: false},
    }),
)
// handle err

err = engine.LoadYAML([]byte(`
version: v1
groups:
  - family: App
    metrics: [app.requests_total]
    charts:
      - id: requests
        title: Requests
        context: requests
        units: requests/s
        dimensions:
          - selector: app.requests_total
            name: requests
`), 1)
// handle err

// 3) Prepare plan from flattened+raw reader, emit it, then commit.
// ReadFlatten() is included even for templates with static dimensions
// because it is required for inferred dimensions, structured-family autogen,
// and is the standard pattern.
attempt, err := engine.PreparePlan(store.Read(metrix.ReadRaw(), metrix.ReadFlatten()))
// handle err
plan := attempt.Plan()
defer attempt.Abort()

err = chartemit.ApplyPlan(api, plan, chartemit.EmitEnv{
    TypeID:      "plugin.job",
    UpdateEvery: 1,
    Plugin:      "example",
    Module:      "example",
    JobName:     "example",
})
// handle err
err = attempt.Commit()
// handle err

PreparePlan Lifecycle

PreparePlan executes a deterministic phase pipeline. Terms like "materialized state" and "route cache" are defined in the Engine State section below.

Phase Summary
Prepare Resolve program/index/cache/materialized state
Validate reader Ensure flattened metadata is available when inferred dimensions are used
Scan Iterate series, filter by success sequence and selector, route to chart/dimension accumulators
Cache retain Prune route cache entries not seen in the latest successful sequence
Lifecycle caps Enforce chart/dimension cap policy
Materialize Emit create/update actions from accumulated state
Expiry Emit removals for stale charts/dimensions
Sort Deterministically sort inferred dimension output

Reader Requirements

Scenario Required reader mode
Static named dimensions only Read(...) is sufficient (no flatten needed)
Inferred dimensions (name and name_from_label omitted) Must use flattened reader metadata (ReadFlatten)
Structured autogen families (Histogram, Summary, StateSet, MeasureSet) Must use flattened reader metadata (ReadFlatten) or they are not visible
Runtime/default ModuleV2 path Read(ReadRaw(), ReadFlatten())

If inferred dimensions are present without flattened reader metadata, PreparePlan returns an explicit error.

Action Semantics

Action Meaning
CreateChartAction Materialize chart instance (with chart metadata and labels)
CreateDimensionAction Materialize dimension for a chart
UpdateChartAction Emit chart values for current cycle; unseen dims, and dims whose value is non-finite (NaN/Inf), become IsEmpty=true (gap, never 0)
RemoveDimensionAction Obsolete one dimension
RemoveChartAction Obsolete one chart

chartemit normalizes emitted action order by phase:

  1. create chart/dimensions
  2. update values
  3. remove dimensions/charts

Routing and Collision Rules

Each metric series is routed to a chart and dimension based on template selectors. The following rules apply when routing conflicts arise:

Rule Behavior
Template vs autogen chart ID collision Template wins; autogen chart is replaced
Cross-template chart ID collision Existing owner keeps ownership; subsequent series are silently ignored (see warning below)
Duplicate dimension observations within build First observed dimension metadata wins; values are reduced (summed)

[!WARNING] Cross-template chart ID collisions cause silent data loss — conflicting series are dropped with no error and no log entry. If metrics are missing, check for duplicate rendered chart IDs across template groups.

Lifecycle Defaults and Policy

Default lifecycle policy when template omits lifecycle:

Policy Default
max_instances 0 (disabled)
expire_after_cycles 5
dimensions.max_dims 0 (disabled)
dimensions.expire_after_cycles 0

Autogen Notes

Topic Behavior
Trigger Unmatched series only when autogen is enabled
Context namespace Autogen context = top-level context_namespace + the full metric name (which includes any SnapshotMeter prefix); empty namespace leaves the bare name. A non-empty meter prefix stacks after context_namespace, so pair context_namespace with SnapshotMeter("") to avoid a doubled prefix
Structured families Autogen has dedicated source builders for flattened Histogram, Summary, StateSet, and MeasureSet families
Metric metadata usage Uses metrix.MetricMeta hints for title/family/unit where allowed
Type ID budget Enforced via AutogenPolicy.MaxTypeIDLen + effective emit type-id prefix (WithEmitTypeIDBudgetPrefix(...))
Lifecycle Autogen applies ExpireAfterSuccessCycles to both chart and dimension expiry (unlike template lifecycle where they default independently)

MeasureSet autogen specifics:

  • chartengine treats MeasureSet as a structured family, similar to StateSet, not as grouped scalar coincidence
  • flattened MeasureSet inputs are expected to carry:
    • SourceKind = MetricKindMeasureSet
    • FlattenRole = FlattenRoleMeasureSetField
    • per-field metric names like <name>_<field>
    • a synthetic reserved field label (measure_field=<field>)
  • the synthetic measure_field label is the authoritative field-identity channel; the per-field metric-name suffix remains for MetricMeta(name) compatibility
  • gauge-like MeasureSet fields autogen with absolute algorithm behavior; counter-like MeasureSet fields autogen with incremental algorithm behavior

Reserved Flattened Label Keys

These label keys are treated specially by chartengine when consuming flattened structured-family or distribution inputs:

Key / Pattern Meaning
le Histogram bucket bound label
quantile Summary quantile label
measure_field MeasureSet field identity label
<metric-name> StateSet special case: the flattened state name is carried under a synthetic label whose key is the base metric name

Notes:

  • le, quantile, and measure_field are static reserved flattened-label keys in chartengine.
  • StateSet is different: it does not use a global static key; it uses the base metric name itself as the synthetic flattened label key.
  • These keys are part of the flatten contract between metrix and chartengine. Reusing them as ordinary user labels on those flattened inputs is not supported.

Runtime Metrics

chartengine self-instruments to a runtime store by default (disable with WithRuntimeStore(nil)).

Family Examples
ChartEngine/Build build success/error/skipped counters, build duration summaries
ChartEngine/Actions action counters by kind
ChartEngine/Series scanned/matched/filtered series counters
ChartEngine/Route Cache hit/miss/entries/prune counters
ChartEngine/Lifecycle removal counters by scope/reason
ChartEngine/Plan gauges for chart instances/inferred dimensions

Engine State

Area Design
Program Immutable compiled IR per revision
Engine state Serialized under Engine.mu for load/build transitions
Route cache Series identity + revision keyed cache; retained by successful sequence, pruned on each build
Materialized state Tracks existing chart/dimension instances for incremental create/update/remove decisions; persists across cycles, resets on template reload
Determinism Sorted chart IDs and inferred dimensions provide stable action ordering