Preserve Feature Flagging configuration source semantics

[leoromanovsky]

Motivation

Complete the Feature Flags configuration migration on top of sealed #11892 without creating phantom billed traffic. New customers default to CDN-backed delivery only after application provider access, while existing users of the legacy experimental enablement variable remain on Remote Configuration during the migration window.

flowchart TD
  Start["Agent starts"] --> Enabled{"DD_FEATURE_FLAGS_ENABLED"}
  Enabled -->|"false"| Disabled["Provider disabled<br/>No RC or CDN traffic"]
  Enabled -->|"true or unset"| Resolve{"Resolve source and legacy state"}
  Resolve -->|"explicit or grandfathered RC"| RC["Subscribe to FFE_FLAGS"]
  Resolve -->|"agentless or new default"| Wait["Wait for application provider access"]
  Resolve -->|"legacy false, invalid, or offline"| Disabled
  Wait --> Access["Application initializes or accesses provider"]
  Access --> CDN["Poll CDN-backed API"]

Loading

Changes

flowchart LR
  Env["Environment configuration"] --> Config["FeatureFlaggingConfig resolver"]
  Config --> Bootstrap["Agent bootstrap gate"]
  Bootstrap -->|"disabled"| Noop["Provider off"]
  Bootstrap -->|"remote_config"| Remote["Eager RC subscription"]
  Bootstrap -->|"agentless"| Lazy["Lazy agentless source"]
  Lazy --> Boundary["Provider access"]
  Boundary --> Poller["UFC HTTP poller"]
  Tests["Resolver, Config, bootstrap,<br/>evaluator, system, and dogfood tests"] -. "verify" .-> Config
  Tests -. "verify lifecycle" .-> Bootstrap

Loading

• Add DD_FEATURE_FLAGS_ENABLED as the canonical provider and delivery kill switch.
• Preserve legacy true as Remote Configuration and legacy false as disabled when no new source is explicit.
• Defer agentless source startup until application code initializes or accesses the Datadog OpenFeature provider.
• Keep explicit and grandfathered Remote Configuration startup eager so the existing FFE_FLAGS subscription is preserved.
• Keep source resolution in the existing Feature Flagging config holder to stay within the agent JAR budget.
• Add resolver, bootstrap gateway, evaluator, system lifecycle, and Config coverage.

Decisions

• DD_FEATURE_FLAGS_ENABLED=false wins over every source and prevents both CDN polling and the FFE_FLAGS subscription.
• An explicit DD_FEATURE_FLAGS_CONFIGURATION_SOURCE wins over the legacy setting.
• With no explicit source, legacy true selects Remote Configuration, legacy false disables Feature Flags, and an absent legacy variable selects agentless.
• Invalid and reserved offline sources fail closed.
• Agentless delivery remains idle until provider access; Remote Configuration remains eager.
• Configuration source remains immutable after initialization.
• This PR is stacked on sealed Add agentless Feature Flagging configuration source #11892 and does not change that parent branch.

Validation

• Full ./gradlew spotlessCheck: passed.
• Focused Feature Flagging config and internal Config tests, plus :dd-java-agent:agent-bootstrap:spotbugsMain: passed.
• :dd-java-agent:verifyAgentJarContents: passed.
• Agent JAR: 34,549,568 bytes, 53,440 bytes below the 34,603,008-byte CI budget.
• system-tests Java source-selection matrix plus generic /ffe/start: 11 passing.
• Local dogfood: no selectors kept startup silent and provider access started agentless delivery; DD_FEATURE_FLAGS_ENABLED=false kept both startup and provider access silent.

[datadog-datadog-prod-us1-2]

🎯 Code Coverage (details)
• Patch Coverage: 56.07%
• Overall Coverage: 57.36% (-0.00%)

_{This comment will be updated automatically if new data arrives.
🔗 Commit SHA: c5bae85 | Docs | Datadog PR Page | Give us feedback!}

[dd-octo-sts]

🟢 Java Benchmark SLOs — All performance SLOs passed

Suite	Status
Startup	🟢 pass

SLO thresholds are defined here based on automatically generated metrics. A warning is raised when results are within 5% of the threshold.

PR vs. master results

Scenario	Candidate	master	Δ (95% CI of mean)
startup:insecure-bank:iast:Agent	13.98 s	13.89 s	[+0.0%; +1.3%] (maybe worse)
startup:insecure-bank:tracing:Agent	12.87 s	12.99 s	[-1.7%; -0.0%] (maybe better)
startup:petclinic:appsec:Agent	16.71 s	16.52 s	[+0.3%; +2.1%] (maybe worse)
startup:petclinic:iast:Agent	16.91 s	16.85 s	[-0.6%; +1.3%] (no difference)
startup:petclinic:profiling:Agent	16.73 s	16.78 s	[-1.3%; +0.7%] (no difference)
startup:petclinic:sca:Agent	17.01 s	16.85 s	[+0.0%; +1.8%] (maybe worse)
startup:petclinic:tracing:Agent	15.54 s	16.26 s	[-8.6%; -0.3%] (maybe better)

Commit: c5bae851 · CI Pipeline · Benchmarking Platform UI

---

Load and DaCapo benchmarks can be triggered manually in the GitLab pipeline. Results will appear in the Benchmarking Platform UI after completion.