Summary
Steeltoe Eureka clients prior to 4.2.0 and 3.4.0 reject valid DataCenterInfo.name values defined by the Java Eureka specification, throwing an exception during registry deserialization.
When this exception is swallowed by the periodic cache refresh task, the local service registry can remain empty or stale, causing service discovery failures in mixed Java/Spring and Steeltoe environments.
Affected context
This vulnerability appears in applications using Steeltoe.Discovery.Eureka as a service discovery client, particularly in environments where Java/Spring Eureka servers and Steeltoe clients coexist.
It is triggered when the Eureka registry contains instances with a DataCenterInfo.name value that Steeltoe does not recognize (such as "Netflix"), even though this value is valid per the Java Eureka specification.
In such deployments, Steeltoe clients repeatedly fail to deserialize the registry payload and silently retain an empty or outdated local service registry cache.
Exploit preconditions
- Attack surface: Service discovery integration between Eureka servers and Steeltoe clients.
- Required access: Ability to register services or modify registry entries in Eureka such that
DataCenterInfo.nameuses a value unsupported by Steeltoe (e.g.,"Netflix"). - Trigger point: Periodic registry cache refresh in the Steeltoe Eureka client where
DataCenterInfo.FromJsonis invoked. - Why exploitation is possible: The client performs strict input validation on
DataCenterInfo.name, throwsArgumentExceptionfor valid but unrecognized values, and the resulting exception is swallowed by the refresh task, preventing proper cache updates (CWE-20, with unbounded error propagation aligning with CWE-400).
Why this happens
The root cause is overly strict and incomplete input validation in DataCenterInfo.FromJson, which only accepts "MyOwn" and "Amazon" and rejects "Netflix", a value allowed by the Java Eureka specification.
This mismatch between accepted values and the protocol specification leads to predictable ArgumentException errors whenever a valid but unsupported DataCenterInfo.name is encountered.
Because the exception is not surfaced or handled in a way that preserves partial state or retries safely, the periodic cache refresh silently fails and leaves the client's registry empty or outdated.
Conceptually, this reflects CWE-20 (Improper Input Validation), combined with CWE-400 (Uncontrolled Resource Consumption) in the sense that repeated failing refresh cycles consume resources without making progress and can effectively degrade the service discovery function.
Potential impact
- Steeltoe clients may retain an empty local registry, causing them to fail to discover any peer services.
- Service discovery can be partially or completely disrupted, leading to cascading failures in microservice calls.
- Traffic may be routed only to stale or previously known instances, increasing error rates and timeouts.
- Operational visibility tools relying on Eureka data may show inconsistent or misleading service topology.
General mitigation guidance
Effective mitigation centers on aligning validation logic with the protocol specification and hardening error handling around registry refresh.
Recommended patterns include:
- Validating
DataCenterInfo.nameagainst the complete set of supported values as defined by the Eureka specification, and treating unknown values as non-fatal with a safe default behavior. - Designing deserialization code to fail gracefully, preserving the last known good cache when encountering unexpected fields rather than wiping the registry.
- Separating parsing and validation concerns so that malformed or unexpected data can be logged and isolated instead of breaking the entire refresh pipeline.
- Implementing robust logging and metrics around registry refresh failures to make misconfiguration and inter-op mismatches visible.
- Avoiding reliance on unbounded, silently failing background tasks; instead, propagate or aggregate errors so operational teams can respond.
How teams detect issues like this
Organizations can identify this vulnerability and similar issues by:
- Reviewing service discovery client code for strict or hard-coded value checks that do not match the upstream protocol or reference implementation.
- Using static analysis or linters to flag unhandled or broadly caught exceptions in background tasks where failures may be silently ignored.
- Adding runtime monitoring around Eureka registry refresh operations, including metrics for successful vs failed refresh cycles and error logging on deserialization failures.
- Incorporating interoperability tests between Java/Spring Eureka servers and Steeltoe clients, covering all documented
DataCenterInfo.namevalues. - Employing automated configuration validation to detect unsupported data center types before deployment.

