RIRRdView on GitHub Evidence behind the beta
Testing and verification.
Recorded protocol, recovery and browser checks, tied to the revision they tested. Passing tests show what works in those fixtures; full IRRd parity remains in development.
Latest published verification
Loading recorded verification…
This is a published evidence snapshot, not a live GitHub status feed. Download the recorded results or inspect CI for the latest runs.
Recorded checks
Each row represents an executed check. Formatting and lint commands have no case counts; skipped cases are shown separately.
| Area | Result | Passed | Failed | Skipped |
|---|
View passed cases by area
Execution coverage
Which lines did we exercise?
No source coverage result has been published yet.
Line coverage measures execution, not correctness or full feature parity. This profile includes test modules within Rust source files; external fixture, harness and dependency files are excluded.
Reproduce the checks
Run the suite locally.
python3 tools/testing/run.py --profile core --out /tmp/rirrd-checks
python3 tools/testing/run.py --profile web --out /tmp/rirrd-web-checks
bash tools/testing/coverage.sh /tmp/rirrd-coverageRequires Rust and Python. Browser checks need Chromium or Chrome; coverage uses the pinned cargo-llvm-cov version documented in the repository.
Full commands & fixture scopeChecks for each push & pull request
Core checks run on every build. Interoperability, router tools and package qualification have dedicated workflows.
Rust & platform checks
Formatting, strict Clippy and locked Rust tests on Linux x86-64, Linux ARM64 and Windows.
Protocol & recovery
TCP, HTTP and GraphQL contracts; source aliases and indexed filters; allocation/IRT authorization, atomic rollback, redaction, restart and backup recovery.
Browser workflows
The bundled console in a real browser against an isolated synthetic registry. Private website and sponsor queue checks run separately.
Coverage & dependencies
Source coverage with native and external tests, dependency auditing and repository hygiene checks.
Parser resilience
Short RPSL and NRTMv3 fuzz runs, plus malformed requests and bounded failure behaviour.
Deployment & comparison
Container smoke checks, with dedicated pinned IRRd comparisons, router integrations and native package workflows.
Workflow configuration and recorded execution are listed separately. GitHub run status has not been verified for this local snapshot. Open the public CI workflow to inspect current runs.
Queries during live mirror commits: the measured development check
Both hosted nodes use native Rust HTTP admission. The qualified beta.4 development build completed 16,086 of 16,086 HTTPS requests during a 70-second run at approximately 230 starts/second, including a real RADB commit. Every successful response contained the complete 54-prefix result; no 429, 502, 503, 504 or transport failure occurred. Recovery passed 5/5.
| Successful response phase | Median | p95 | p99 |
|---|---|---|---|
| Total | 17.68 ms | 22.98 ms | 26.04 ms |
| After TLS | 2.53 ms | 4.31 ms | 6.09 ms |
The driver ran on the hosting VM using pinned IPv6 with certificate and hostname verification. This is a warm-cache, self-routed HTTPS regression check, not an Internet benchmark or maximum-capacity claim. It uses a different workload from the large-query and historical comparison measurements.
Read the exact scope, artifact and complete-answer evidence →
Passing checks are part of the picture.
RIRRd remains beta. Full authentication and account workflows, remaining class and policy behaviour, broader mirroring interoperability and release qualification still need work. Passing these checks establishes the behaviour exercised by their fixtures.