Frequently Asked Questions

The Probe intercepts every source of non-determinism at I/O boundaries — Date.now(), Math.random(), HTTP responses, database results, DNS lookups, file reads, and timer callbacks. During replay, all of these are served from the recording. Your application code runs completely unmodified; it simply sees the same data it saw in production.

Less than 1% CPU and approximately 2MB/minute per service at 100 req/min. The Probe uses a lock-free spill buffer — if the Collector is unreachable, events are buffered in memory/disk locally and flushed automatically when connectivity returns. Pro users can enable smart sampling to record only 1-in-N requests, reducing overhead to essentially zero.

Any framework built on Node.js I/O primitives works out-of-the-box: Express, Fastify, Koa, NestJS, Hapi, and raw http.createServer. Database support includes pg (PostgreSQL), mysql2, and any library wrapping Node's net.Socket. ORMs — Prisma, Sequelize, TypeORM — are all covered automatically.

Yes — that's the Pro tier's core capability. The Collector correlates events across services using HLC (Hybrid Logical Clock) timestamps and the original request's propagation chain. During replay, all services execute in lock-step, preserving the exact inter-service call order and timing relationships from production.

Replay.io and rr are browser/process-level recorders for single-process debugging. ERGENEKON is a distributed systems recorder — it captures I/O interactions across an arbitrary number of networked services and replays them in lock-step.

Think of rr as a microscope. ERGENEKON is an MRI machine for your entire production stack.

The Collector runs on your own infrastructure — nothing leaves your network unless you choose otherwise. All recordings stay on your servers. Pro users get deep PII redaction rules (regex patterns, JSON path selectors) that scrub sensitive fields before events are written to disk. Field-level encryption is also configurable.

Licenses are Ed25519-signed JWT tokens verified entirely locally. No license server call is needed at runtime — not even on startup. The private key never leaves our servers; you receive a signed token that your local installation verifies cryptographically. Air-gapped environments are fully supported at the Enterprise tier.

We're in launch mode. We want real engineers at real companies to integrate ERGENEKON, stress-test it in production, and tell us what's broken. In exchange, you get full Pro/Enterprise access at zero cost.

Community: 24 hours · Pro: 30 days · Enterprise: Unlimited.

Retention is enforced by the Collector's built-in TTL eviction. You can pin any session with ergenekon session pin <id> to prevent eviction regardless of tier.

Yes. The Collector ships as a stateless binary that persists sessions to a mounted volume — drop it into any container. Official Docker Compose and Helm chart examples are in the repo under /deploy.

For Kubernetes, we recommend a DaemonSet per node-pool with a shared PVC for session storage. Enterprise tier includes a production-hardened Helm chart with SLA support.

Q3 2026: Rust probe (sub-0.1% overhead), Go probe, Python probe (Django/FastAPI).
Q4 2026: AI-powered root cause analysis — ERGENEKON surfaces the exact line of code responsible for a production bug.
2027: Predictive recording — automatically capture sessions matching anomaly signatures before a bug even manifests.

ERGENEKON is released under Business Source License 1.1 (BSL-1.1). You can use it freely in production for any purpose. The only restriction: you cannot offer ERGENEKON itself as a commercial hosted service (i.e., re-selling our tool as SaaS).

Email us at hello@ergenekon.dev — we respond to every email within 48 hours. For bugs and feature requests, open a GitHub Issue. For security-related disclosures, please use security@ergenekon.dev.