TopGit
GitHub Repo Review

Netflix Hystrix: Maintenance-Mode Circuit Breaker for Java

Netflix/Hystrix
NTopGit review image for Netflix/Hystrix
Review by Topgit.dev for Netflix/Hystrix, with GitHub repository stats and README context.
Quick verdict

Netflix Hystrix is a battle-tested circuit breaker library that Netflix itself has stopped developing. It still does what it always did — thread and semaphore isolation that stops one flaky dependency from taking down a whole service. Starting a brand-new project on it is a harder sell: the README tells you to use Resilience4j instead, and there's no one on the other end of a GitHub issue anymore.

Stars
★ 24.5k
Forks
⑂ 4.7k
Language
Java
License
See repository
Topic
—
Updated
Dec 2025
Homepage
—
GitHub

What Is Netflix Hystrix?

Netflix Hystrix is a Java library that wraps calls to remote services, other systems, or third-party libraries inside a HystrixCommand so a slow or failing dependency can't cascade into a full outage. Each command runs inside its own thread pool or semaphore, trips a circuit breaker when failures pile up, and falls back to a default response instead of hanging. Netflix built it to keep requests moving even when downstream calls fail.

Core Features of Hystrix

  • ✓Thread and semaphore isolation — each HystrixCommand runs in its own thread pool (or a semaphore for lighter calls) so a stuck dependency can't exhaust the caller's own threads.
  • ✓Circuit breakers that trip automatically once a command's failure rate crosses its configured threshold, then periodically let a test request through to check recovery.
  • ✓Fallback methods — a command can define a fallback that returns a default or cached value instead of propagating the failure up the call stack.
  • ✓Request collapsing, which batches multiple calls made in a short window into a single backend request.
  • ✓Request caching scoped to a single request context, so duplicate calls for the same data inside one request hit the cache instead of the network.
  • ✓Realtime property changes — timeout, isolation, and threshold settings can be updated live across a fleet without redeploying.
  • ✓Three execution styles per command: execute() for a blocking call, queue() for a Future, and observe() for an RxJava Observable.
How this repository's GitHub stars have grown over time. Source: star-history.com.View the star history ↗

Strengths

  • ✓The isolation model is simple to reason about: one command, one thread pool or semaphore, one circuit breaker — failures stay contained to the dependency that caused them.
  • ✓Fallback methods make graceful degradation an explicit part of the code instead of an afterthought bolted on with try/catch.
  • ✓It was built and hardened inside Netflix's own services, so the failure scenarios it isolates against — slow third-party calls, cascading timeouts — are real ones, not textbook edge cases.
  • ✓Maven Central distribution means adding it is a single dependency line, and existing Hystrix code keeps running exactly as before.

Hystrix Is in Maintenance Mode

  • △Netflix put Hystrix in maintenance mode: no more issue triage, pull request review, or new releases from the Netflix team.
  • △The last release, 1.5.18, is a final bookkeeping release to match the version actually used internally (1.5.11) — not a sign of continued feature work.
  • △The README itself recommends Resilience4j for new internal projects. That's an odd position for a library to be in.
  • △The dashboard for watching circuit breakers in real time was pulled out of this repo entirely and lives in a separate Netflix-Skunkworks project now, so monitoring isn't included out of the box.
  • △Who takes over the project is an open question — the README invites anyone interested to email [email protected], but nothing here documents that handoff having happened.

Hystrix Alternatives to Consider

Resilience4j — the library Hystrix's own README tells new projects to use instead; lighter-weight, still actively maintained, and built around Java 8 functional interfaces rather than Hystrix's command classes.Sentinel — Alibaba's traffic-control and circuit-breaking library, actively maintained, with a similar isolation and circuit-breaker model but its own dashboard and rule system.

Frequently Asked Questions

Is Netflix Hystrix still actively maintained?

Netflix Hystrix is not under active development anymore — it moved into maintenance mode, meaning Netflix no longer reviews issues, merges pull requests, or ships new releases. The final release is 1.5.18, published to align Maven Central with the last version Netflix ran internally (1.5.11).

Is Hystrix free for commercial use?

Hystrix is free for commercial use. The GitHub repo licenses Hystrix under the Apache License, Version 2.0, which permits commercial use, modification, and redistribution without paying royalties or opening up your own source code in return.

Should I use Hystrix or Resilience4j for new projects?

For a new project, choose Resilience4j over Hystrix — the Hystrix README itself says Netflix now recommends Resilience4j for new internal work and is only keeping Hystrix around for existing applications. Picking Hystrix today means building on a library its own maintainers have moved away from.

What does Hystrix maintenance mode mean for existing applications?

Maintenance mode means existing applications that already use Hystrix keep working exactly as before — the isolation, circuit breaker, and fallback behavior hasn't changed. It just means Netflix won't fix new bugs, add features, or merge community pull requests going forward, so any future issue is yours to fix or fork.

How do circuit breakers work in Hystrix?

Hystrix wraps a remote call inside a HystrixCommand and tracks its recent success and failure rate. Once failures cross the command's configured threshold, the circuit breaker trips, stops sending requests to that dependency, and routes calls to a fallback instead. It periodically lets a single test request through to check whether the dependency has recovered, closing the circuit again if it has.

Can I migrate from Hystrix to Resilience4j?

There's no built-in migration tool — the Hystrix README doesn't document one. Moving to Resilience4j means rewriting each HystrixCommand's isolation, fallback, and circuit-breaker logic with Resilience4j's own API, since the two libraries aren't drop-in compatible. Plan on a manual, command-by-command rewrite.

The problem it solves

Netflix Hystrix targets a narrower problem than "distributed systems are hard": a single flaky remote call — a slow payment API, a stalled recommendation service — that, left unisolated, ties up caller threads until the whole request pipeline backs up behind it. Hystrix's answer is to wrap each remote call in its own bulkhead so one bad dependency degrades gracefully instead of taking the rest of the request down with it.

Best use cases

  • •Wrapping a call to a third-party payment or shipping API where an outage on their end shouldn't take your checkout flow down with it.
  • •Isolating a recommendation or personalization service that's nice-to-have — if it's slow, fall back to a generic response instead of blocking the page.
  • •Batching many small lookups, like user account fetches, into one call via request collapsing instead of hammering a backend service per item.
  • •Maintaining an existing Netflix-era microservices codebase that already depends on Hystrix, where a rewrite to Resilience4j isn't worth the churn yet.

How to install / try

Add the Maven dependency com.netflix.hystrix:hystrix-core with the version you want (the README points to Maven Central for the exact coordinates), or the equivalent Ivy dependency. To build from source instead: clone the repo and run ./gradlew build — Java 6 or later is required. There's no separate install step beyond pulling the JAR through your build tool.

How to use

You subclass HystrixCommand, put the protected logic inside a run() method, and give the command a group key in the constructor. Calling execute() runs it synchronously and returns the result or the fallback, queue() returns a Future, and observe() returns an RxJava Observable for reactive callers. The README's own example wraps a trivial "Hello, name" call this way, and the hystrix-examples module has fuller demos, including one you can run with ./gradlew runDemo to watch circuit breakers trip and recover against simulated failures in real time.

Who should try it — and who should skip

Teams already running Hystrix in a live Netflix-era Java microservices stack should keep it — it isn't broken, and the isolation guarantees haven't changed. Teams starting something new should skip it and look at Resilience4j instead, the project Netflix itself now points people toward. Anyone who needs an actively maintained project to file issues against or expect new releases from is in the wrong place — that's the whole point of maintenance mode.

Source & attribution

Based on the Netflix/Hystrix GitHub repository and its README (https://github.com/Netflix/Hystrix).

GitHub data · last synced Aug 15, 2026Reviewed by Henry
← Back to TopGit

Still deciding about Hystrix?

One click hands the question to an AI along with this page — see what it says about Hystrix.

GitHub