Skip to content

About

Ruby/Rails integration and release lab: reliable partner publishing, signed deployments, failure diagnosis and recovery.

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Repository files navigation

MESH · Integration & Release Lab

Verification

A Ruby/Rails engineering showcase for external integrations, release automation and recovery.

MESH starts as a creator marketplace: brief → proposal → agreement → versioned delivery → acceptance. Its Publishing lab demonstrates what happens when a partner accepts a release but the response is lost: retain the operation, diagnose uncertainty and reconcile without publishing twice.

Watch the short demo · Screenshot walkthrough · Verified results

Real Publishing workflow: package validation, uncertain release and safe recovery

Actual application recording with English captions. The application UI is in Russian. The public presentation is static; it does not expose the backend or real partner credentials.

What the project demonstrates

Engineering problem Implementation
Repeated or concurrent commands Idempotency keys, PostgreSQL constraints, row locks and immutable agreements
Uncertain external effects Durable operation identity, lookup-only reconciliation, lease fencing and signed callback deduplication
Unsafe or stale release inputs Private digest-bound ZIP artifacts, bounded validation, real ClamAV, policy/configuration fingerprint
Reliable delivery Outbox, Solid Queue, durable dispatcher and metrics that include unfinished work
Trusted release and recovery Signed image provenance, deployment by digest, migration/runtime DB roles, rollback journal and encrypted restore

Architecture: eight-domain modular Rails monolith. Publishing execution has framework-independent Domain/Application layers with explicit ports and PostgreSQL/HTTP/storage adapters. Code map and tradeoffs.

Stack: Ruby / Rails · PostgreSQL · Solid Queue · Next.js / TypeScript · Node.js partner simulator · Docker · GitHub Actions · Ansible · Terraform · Helm / Kubernetes · Prometheus / Grafana / Alertmanager.

Executed evidence

The accepted application release is 3d849af. Later documentation commits retain that release identity.

Verification Result Evidence
Ruby and browser 153 Ruby examples in each of development and native Alpine; 13 browser scenarios; no failures, pending, skipped or flaky scenarios CI
Exact-image security Five runtime digests scanned; zero HIGH/CRITICAL findings, including unfixed; four signed build outputs Release
Ubuntu deployment Ansible second apply changed=0, mixed load, rejected release rollback and eight SIGKILL recovery points Deployment
Kubernetes Terraform drift repair, real NetworkPolicy controls, readiness/liveness and rollout/rollback Cluster exercise
Clean demo preparation Fresh Ubuntu checkout; actual npm run demo; all 17 stages and 4 browser checks passed Cold preparation
Separate-VM recovery Primary/queue/private artifacts restored; API/workers resumed; no repeated external publication Recovery

The clean-run preparation and gallery distinguish fresh checkout verification from an existing local preview. Reports retain source revisions, workload/fault scope, original artifact digests and per-file hashes. Verify the saved catalogue with npm run test:partner-support:evidence. A passing scan is a dated observation, not a permanent security guarantee.

Run locally

Requirements: Node.js 22/npm, Docker with Linux containers, registry access, Microsoft Edge on Windows (or the matching Playwright Chromium runtime on Linux) and enough memory for the application plus ClamAV. Prepare before the interview; the command includes builds and can take several minutes.

npm run demo

Open Publishing, marketplace and Grafana. Synthetic operator: ops@mesh.local / MeshDemo2026! — lab credentials only.

npm run demo:publishing       # real partner, lost response, reconciliation and rollback
npm run demo:partner-support # disposable peer: credential rotation, fencing and uncertainty
npm run demo:incident        # worker outage, delivered alert and once-only recovery

Setup and environment guide · 15-minute technical walkthrough · Detailed marketplace guide (RU)

Scope and limits

This is a reproducible engineering lab, not a claim of operating a commercial gaming platform. Uploaded code is never executed; payments and partners are simulators. Hosted environments are ephemeral, Kubernetes is single-node, and restore uses a quiescent snapshot with a small fixture. Multi-node HA, online WAL/PITR, large-restore throughput, permanent backup retention and a long production SLO window are not demonstrated.

Read deeper

About

Ruby/Rails integration and release lab: reliable partner publishing, signed deployments, failure diagnosis and recovery.

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages