阅读中文版本

Build a Service Discovery Registry

Service Discovery Registry — a complete walkthrough of service discovery, covering instance registry, heartbeat tracker, ttl evictor, and testing.

service_discovery_registry/
├── src/index.ts
├── src/instance-registry.ts
├── src/heartbeat-tracker.ts
├── tests/service_discovery_registry.test.ts
└── package.json

// Instance registry: service discovery core
export function instanceregistry(input: InstanceregistryInput): InstanceregistryResult {
  // Build the instance registry that anchors service discovery and defines the core data shape.
  return processInstanceregistry(input);
}

// Heartbeat tracker — handles service discovery Add the heartbeat tracker so
export class Heartbeattracker {
  constructor(private opts: HeartbeattrackerOptions) {}
  async run(ctx: Context): Promise<HeartbeattrackerOutput> {
    return this.execute(ctx);
  }
}

// TTL evictor: service discovery stage
export const ttlevictor = (next: Next) => async (req: Request) => {
  const out = await transform(req);
  return next(out);
};

// Watch notifier for service discovery
export async function watchnotifier(stream: AsyncIterable<Chunk>): Promise<Aggregate> {
  let acc = init();
  for await (const c of stream) acc = merge(acc, c);
  return acc;
}

// Query endpoint: service discovery finalization
export function queryendpoint(state: State): Result {
  validate(state);
  return finalize(state);
}

Project overview

This project builds a service discovery system whose reason for existing is maintaining a live directory of service instances that register, heartbeat, and deregister. The temptation in service discovery is to bolt the interesting part onto a working prototype and discover the hard cases in production; this build inverts that order and treats a network partition that splits registrations across a majority and minority that both serve stale instance lists as the first thing designed, not the last.

You will implement Instance registry, Heartbeat tracker, TTL evictor, Watch notifier, and Query endpoint, each as a module with one responsibility and a test that proves it in isolation. The service discovery pipeline is then the wiring that composes them, and the integration test is where the composition is shown to survive a network partition that splits registrations across a majority and minority that both serve stale instance lists.

The walkthrough is written for a developer who has used a service discovery system but has not had to make one honest. It favors explicit data shapes over clever abstractions, and it names the failure each design choice prevents rather than presenting the choice as a style preference.

Each phase ends with something observable — a instance registry that returns the right shape, a heartbeat tracker that rejects bad input, an integration test that goes green — so you can stop at any point and still have a service discovery system that works for the cases covered so far.

What you will build

The finished service discovery system exposes five capabilities, each with a focused test:

• Instance registry
• Heartbeat tracker
• TTL evictor
• Watch notifier
• Query endpoint

Instance registry is the foundation that every later stage assumes; Heartbeat tracker is where service discovery correctness lives; TTL evictor handles a network partition that splits registrations across a majority and minority that both serve stale instance lists; Watch notifier makes the system observable; and Query endpoint is the graceful-degradation layer that keeps overload from becoming corruption.

The build keeps the service discovery logic in ordinary modules you can read in one sitting, and pulls in dependencies only for the parts that are genuinely tedious to hand-roll. That keeps the diff between the tutorial and a real codebase small, and keeps the failure modes visible in code rather than hidden behind a framework.

Architecture and data model

The system is organized around Instance registry, Heartbeat tracker, and TTL evictor, with Watch notifier tying them together. Data flows as a map of service names to instance lists, each instance evicted when its heartbeat TTL expires without renewal.

Each component is a module with a pure interface where practical: the same input produces the same output, and side effects are pushed to the edges. That discipline is what makes the phases independently testable, because a test for TTL evictor does not need to stand up Instance registry and Heartbeat tracker to exercise the interesting branch.

The data model is deliberately explicit. A map of service names to instance lists, each instance evicted when its heartbeat TTL expires without renewal Keeping the shape written down — in a schema, a type, or a fixture — means a change to the shape is a visible diff rather than a silent runtime shift, and it means the integration test can assert against a concrete shape instead of guessing.

Configuration and code are kept apart: the configuration says what service discovery should do, the modules say how. A reviewer can change a threshold or a policy without reading the implementation, which keeps the risky changes small and lets the diff stand as the review.

The failure path through TTL evictor is designed before the happy path is finished, because a network partition that splits registrations across a majority and minority that both serve stale instance lists is the case that distinguishes a service discovery system from a demo. Every component declares what it does when its inputs are wrong, and the integration test asserts each of those declarations.

Prerequisites and dependencies

Have these ready before you install anything, so the service discovery setup is not blocked by a missing tool:

• Node.js 20 LTS or later
• Familiarity with service discovery fundamentals
• A local environment to run the test suite

The runtime dependencies are chosen to keep the service discovery datapath legible:

• fastify (^5.0.0) — Registry API
• redis (^4.7.0) — Shared registry state

# setup for service-discovery-registry
pnpm init
pnpm add fastify redis
pnpm add -D typescript tsx vitest

Versions are pinned to a major range so the service discovery install is reproducible; commit the lockfile. The instance registry reads its operational parameters from the environment and validates them at startup, so a missing value fails the boot rather than the first service discovery request.

Directory structure

Keep each service discovery component in its own file so it can be tested without starting the others:

service_discovery_registry/
├── src/index.ts
├── src/instance-registry.ts
├── src/heartbeat-tracker.ts
├── tests/service_discovery_registry.test.ts
└── package.json

Phase 1: Instance registry

Build the instance registry that anchors service discovery and defines the core data shape.

// Instance registry: service discovery core
export function instanceregistry(input: InstanceregistryInput): InstanceregistryResult {
  // Build the instance registry that anchors service discovery and defines the core data shape.
  return processInstanceregistry(input);
}

After this phase the instance registry works in isolation and is covered by a unit test. The instance registry defines the data shape that every later stage assumes, so getting it right first means the later phases compile against a stable contract.

Treat the instance registry as the single source of truth for the service discovery data shape. Validate inputs at the boundary and reject early, so a malformed value never reaches Heartbeat tracker or TTL evictor where its effect would be harder to localize.

Phase 2: Heartbeat tracker

Add the heartbeat tracker so service discovery can maintaining a live directory correctly.

// Heartbeat tracker — handles service discovery Add the heartbeat tracker so
export class Heartbeattracker {
  constructor(private opts: HeartbeattrackerOptions) {}
  async run(ctx: Context): Promise<HeartbeattrackerOutput> {
    return this.execute(ctx);
  }
}

After this phase the heartbeat tracker works in isolation and is covered by a unit test. This is where service discovery correctness lives: the heartbeat tracker is what makes maintaining a live directory of service instances that register, heartbeat, and deregister true, and its test is the proof.

Keep the heartbeat tracker pure where you can. A function that takes an input and returns an output is trivial to test and impossible to mock wrong; a class with hidden state is the opposite. Push side effects to the caller and keep the heartbeat tracker about the transformation.

Phase 3: TTL evictor

Wire the ttl evictor to handle a network partition that without losing data.

// TTL evictor: service discovery stage
export const ttlevictor = (next: Next) => async (req: Request) => {
  const out = await transform(req);
  return next(out);
};

After this phase the ttl evictor works in isolation and is covered by a unit test. The ttl evictor is the difference between a service discovery system that works on the happy path and one that survives a network partition that splits registrations across a majority and minority that both serve stale instance lists.

Design the failure before the success. Decide what ttl evictor does when a network partition that splits registrations across a majority and minority that both serve stale instance lists: reject, retry, shed, or degrade — and write that test first. A service discovery system whose failure behavior is unspecified will pick one at random under pressure, and it is usually the wrong one.

Phase 4: Watch notifier

Integrate the watch notifier to make service discovery observable and testable end to end.

// Watch notifier for service discovery
export async function watchnotifier(stream: AsyncIterable<Chunk>): Promise<Aggregate> {
  let acc = init();
  for await (const c of stream) acc = merge(acc, c);
  return acc;
}

After this phase the watch notifier works in isolation and is covered by a unit test. Observability is not a finishing touch; it is how you know the service discovery system is doing what it claims in production.

Emit one structured event per meaningful action, with enough context to localize a failure without a second round-trip. The watch notifier is what turns “something is slow” into “the heartbeat tracker is slow for inputs matching this shape,” which is the difference between an alert and an investigation.

Phase 5: Query endpoint

Finalize the query endpoint so the service discovery system degrades gracefully under overload.

// Query endpoint: service discovery finalization
export function queryendpoint(state: State): Result {
  validate(state);
  return finalize(state);
}

After this phase the query endpoint works in isolation and is covered by a unit test. The query endpoint is what keeps overload from becoming corruption: when the service discovery system is asked for more than it can do, it degrades in a defined way instead of failing in an undefined one.

Pick a degradation strategy for query endpoint — shed, queue, or backpressure — and make it explicit in code and in a test. Leaving the behavior to chance is how a service discovery traffic spike becomes a data loss incident.

Testing

Each phase has a focused test: instance registry, heartbeat tracker, and ttl evictor are exercised in isolation, then an integration test drives the full service discovery pipeline and asserts the a network partition that case is handled.

Start with unit tests that pin each component’s contract: the instance registry returns the documented shape, the heartbeat tracker transforms correctly, and the ttl evictor fails in the documented way. These tests are fast and they localize regressions to the component that broke.

Then add an integration test that composes the components and drives a realistic service discovery scenario end to end. Assert the happy path first, then deliberately trigger a network partition that splits registrations across a majority and minority that both serve stale instance lists and assert the system degrades rather than corrupts. Because the components are pure where practical, the integration test stays deterministic rather than flaky.

A service discovery suite must also prove the negative: bad input is rejected at the boundary, overload is shed rather than queued indefinitely, and a partial failure does not leave the system in an inconsistent state. These are the tests that catch the regressions a happy-path suite never would.

The entire service discovery suite runs with no external service; every dependency is either in-process or stubbed, so every case is reproducible in CI. That is what lets the ttl evictor failure-mode assertions live as tests rather than as a manual checklist that drifts out of date.

# run the service discovery test suite
pnpm vitest run

Deployment and operations

Run at least two service discovery instances behind a load balancer and share state in Redis. Set resource limits so service discovery overload surfaces as backpressure rather than failure.

Run at least two stateless instances behind a load balancer, and move any shared state into a shared store so the service discovery behavior is global rather than per instance. Size limits so the system sheds load before it runs out of memory, because a service discovery system that OOMs under load fails in a way that is hard to reason about.

Drain on deploy: stop accepting new work, let in-flight service discovery operations finish, then exit. Because the system holds no state that cannot be reconstructed, a rollout is a matter of cycling instances one at a time while watching the error rate.

When a dependency is unavailable, decide explicitly whether the service discovery system fails closed or degrades. Leaving that choice to chance is how a dependency outage becomes a silent data loss, so make the choice visible in code and in a test.

Security considerations

Validate all service discovery inputs at the boundary, reject oversized payloads before buffering them, and keep secrets in the environment rather than the repository.

The service discovery system is a choke point, so it is the right place to reject malformed input before it reaches heartbeat tracker or ttl evictor. Validate the shape and the size at the edge, and let the inner components assume what they receive is well-formed.

Never log secrets or full payloads; log enough to localize a failure without recording data that should not be on disk. Treat any external call as untrusted and validate its response before using it, because a service discovery system that trusts its inputs blindly is one malformed response away from corruption.

Keep dependencies pinned and run an audit in CI. A transitive dependency with a known vulnerability is still a vulnerability in your service discovery system, and the lockfile is what makes the audit meaningful.

Observability

Emit one structured event per service discovery operation with the inputs’ shape, the outcome, and a correlation id. Track error rate and latency per component and alert on a step change rather than a single spike.

The correlation id is the thread that ties a user-visible symptom to the service discovery decision and to the component log entry for the same operation. Without it, an investigation starts from a timestamp and a guess.

Watch the ttl evictor failure rate separately from the overall error rate. A jump in ttl evictor failures usually means a network partition that splits registrations across a majority and minority that both serve stale instance lists, not a bug in heartbeat tracker, and the two have different owners and different fixes.

Chart the share of service discovery operations that shed, queue, or succeed. A change in the mix points at the cause faster than an aggregate latency can, because each class has a different root.

Troubleshooting

A network partition that splits registrations across a majority and minority that both serve stale instance lists. Reproduce with a minimal fixture, then add a guard test so the service discovery system rejects the condition loudly.

Performance regresses under service discovery load. Profile with the sampling analyzer, locate the hot path in heartbeat tracker, and add a targeted benchmark to lock the regression.

Keep the service discovery dashboards separate from dependency dashboards. When an operation fails, the correlation id lets you decide in seconds whether the fault is instance registry, heartbeat tracker, or ttl evictor, instead of guessing from a timeout.

Limitations and next steps

• Single-region; multi-region replication is out of scope.
• The instance registry is in-process and not horizontally shared by default.
• No auth layer; assume deployment behind an edge that enforces identity.

The service discovery system is single-region by design; extending to multi-region means making instance registry shareable across regions, which is a real change to the consistency model rather than a config flag. The ttl evictor is in-process, which is correct for one instance but must move to a shared store before scaling out.

There is no authentication or authorization layer; the system assumes it is deployed behind an edge that enforces identity. Adding one is straightforward but out of scope for this build, and doing it here would distract from the service discovery correctness that is the point.

The natural next step, once the service discovery pipeline is trusted in production, is to add a short-TTL cache for idempotent reads and a richer watch notifier that exports histograms rather than counters. Both are additive and neither changes the contracts this build establishes.