● Beta — KubeEra is in active development. Profiles are live; AI-perception measurement is rolling out.
KubeEra
C

Chaos Mesh

IncubatingObservability
Open source · CNCF · CNCF Incubating · Apache-2.0

Chaos Mesh is a chaos engineering platform for Kubernetes.

updated 2026-07-01

What it is

Chaos Mesh is a chaos engineering platform for Kubernetes. It lets you inject faults — pod kills, network latency/partition, IO delays, kernel panics, DNS errors, clock skew — into running workloads to see what actually breaks before production does it for you. It ships as a set of CRDs and controllers, so experiments are defined declaratively and run inside the cluster.

Worth flagging: the category tag here is "Observability and Analysis," but that's not really what Chaos Mesh does. It doesn't observe anything by itself — it breaks things on purpose. You'd pair it with Prometheus/Grafana or similar to observe the blast radius of the chaos it creates.

Who builds it and why

Originally built by PingCAP to stress-test TiDB's distributed failure modes, then donated to CNCF. It's now a CNCF incubating project with 221 contributors and active commits as recently as mid-2026 — this isn't an abandoned donation coasting on its CNCF badge. Founding date isn't publicly documented, but the project has been through multiple major version lines (currently v2.8.3), indicating sustained investment rather than a one-off release-and-forget.

Production readiness signal

CNCF incubating status means it's cleared a bar for adoption diversity and governance, but hasn't hit graduated maturity. 7,776 GitHub stars and 221 contributors is a healthy, active mid-tier CNCF project — not as large as top-tier projects like Argo or Istio, but not a niche experiment either. Recent commit activity suggests it's maintained, not stagnant. Apache-2.0 license, Go codebase (consistent with the rest of the K8s ecosystem, easy to build/extend/debug alongside your other tooling).

The honest caveat: chaos engineering tools are inherently risky by design — you're running something whose job is to break your cluster. Production readiness here means "the tool reliably does what it says," not "safe to point at prod on day one." Treat any chaos tool with the same operational caution you'd give a database migration script.

Who should use this

  • Teams running distributed, stateful, or multi-service systems where failure modes are non-obvious (databases, service meshes, event-driven architectures).
  • Platform/SRE teams building a formal chaos engineering practice with GitOps-style, repeatable, version-controlled experiments (as opposed to ad-hoc kubectl delete pod cowboy testing).
  • Orgs already invested in Kubernetes CRDs/controllers as the pattern for infra tooling — this fits that model natively.
  • Anyone needing fine-grained fault injection (network partition, IO delay, clock skew) beyond what basic pod-kill scripts give you.

Who should NOT use this

  • Small teams running a handful of stateless services with simple failure modes — the operational overhead of running Chaos Mesh's controllers and experiment CRDs isn't worth it; a shell script killing pods on a cron job does the job.
  • Teams without solid observability already in place. Chaos Mesh tells you nothing about what broke — if you don't have monitoring/tracing/logging mature enough to observe the blast radius, you're just causing outages with extra steps.
  • Regulated or high-stakes production environments without a mature staging/canary process — running fault injection without a safe environment to test in first is how you cause the incident you were trying to prevent.
  • Teams wanting a fully managed, vendor-supported chaos-as-a-service offering with SLAs — Chaos Mesh is self-hosted and self-supported.

Alternatives

  • Litmus Chaos — another CNCF chaos engineering project, similar CRD-based model, arguably broader hub/experiment marketplace.
  • Gremlin — commercial, managed chaos engineering SaaS with a polished UI and support contract, if you want vendor accountability instead of self-hosting.
  • AWS Fault Injection Simulator — if you're all-in on AWS and want fault injection tightly integrated with AWS-native services rather than Kubernetes-centric.

Pricing

Fully open source, Apache-2.0. No paid tier, no commercial entity behind it selling support contracts as far as publicly available information shows. You run it, you own the operational risk and the maintenance burden.

Frequently asked

What is Chaos Mesh?+
Chaos Mesh is a chaos engineering platform for Kubernetes. It lets you inject faults — pod kills, network latency/partition, IO delays, kernel panics, DNS errors, clock skew — into running workloads to see what actually breaks before production does it for you.
Who builds Chaos Mesh?+
Originally built by PingCAP to stress-test TiDB's distributed failure modes, then donated to CNCF. It's now a CNCF incubating project with 221 contributors and active commits as recently as mid-2026 — this isn't an abandoned donation coasting on its CNCF badge.
Is Chaos Mesh production ready?+
CNCF incubating status means it's cleared a bar for adoption diversity and governance, but hasn't hit graduated maturity. 7,776 GitHub stars and 221 contributors is a healthy, active mid-tier CNCF project — not as large as top-tier projects like Argo or Istio, but not a niche experiment either.
Who should use Chaos Mesh?+
Teams running distributed, stateful, or multi-service systems where failure modes are non-obvious (databases, service meshes, event-driven architectures).
Who should not use Chaos Mesh?+
Small teams running a handful of stateless services with simple failure modes — the operational overhead of running Chaos Mesh's controllers and experiment CRDs isn't worth it; a shell script killing pods on a cron job does the job. Teams without solid observability already in place.
What are the alternatives to Chaos Mesh?+
Litmus Chaos — another CNCF chaos engineering project, similar CRD-based model, arguably broader hub/experiment marketplace. Gremlin — commercial, managed chaos engineering SaaS with a polished UI and support contract, if you want vendor accountability instead of self-hosting.
How much does Chaos Mesh cost?+
Fully open source, Apache-2.0. No paid tier, no commercial entity behind it selling support contracts as far as publicly available information shows. You run it, you own the operational risk and the maintenance burden.