015_Broker_Health_Probes

Broker Health Probes

Wolverine exposes an optional, transport-implemented contract for monitoring the underlying broker connection. It is intentionally distinct from the existing WolverineTransportHealthCheck (which feeds the ASP.NET Core health-check pipeline) — the broker probe is meant for interactive monitoring layers like CritterWatch that want to render broker connectivity at a glance.

The contract

public interface IBrokerHealthProbe
{
    Task<BrokerHealthSnapshot> ProbeAsync(CancellationToken ct);
}
 
public record BrokerHealthSnapshot(
    Uri TransportUri,
    string TransportType,
    BrokerHealthStatus Status,           // Unknown / Healthy / Degraded / Unhealthy
    string? Description,
    string? CertificateExpiry,           // ISO-8601 if TLS configured
    int ReconnectAttempts,
    DateTimeOffset LastSuccessfulAt);

Probes are non-destructive: they do not reconnect, bounce the connection, or mutate transport state. They inspect the connection the transport already holds and return a snapshot. If the connection is currently down, the probe returns Unhealthy rather than throwing.

Discovering probes at runtime

Every ITransport registered with Wolverine is iterable via runtime.Options.Transports. Probes are discovered by filtering that collection:

var snapshots = await Task.WhenAll(
    runtime.Options.Transports
        .OfType<IBrokerHealthProbe>()
        .Select(p => p.ProbeAsync(cancellationToken)));

A transport that does not implement IBrokerHealthProbe simply doesn’t contribute a snapshot.

Status semantics

Status	When it’s reported
Unknown	Transport hasn’t connected yet (host hasn’t started, or disabled).
Healthy	Connection open, no recent flap.
Degraded	Connection open, but there’s been a recent reconnect.
Unhealthy	Connection currently down.

ReconnectAttempts counts genuine recovery events — it is not incremented by the initial connect at host startup. LastSuccessfulAt is the timestamp of the most recent successful connect (initial or recovered).

Transport support

Transport	Probe support
RabbitMQ	Yes (since this release)
Azure Service Bus	Pending
Amazon SQS / SNS	Pending
Apache Kafka	Pending
Apache Pulsar	Pending

The follow-up transports will pattern-match the RabbitMQ implementation as their underlying client SDKs are surveyed.

RabbitMQ specifics

For RabbitMQ, the probe inspects:

• IConnection.IsOpen for both the listening and sending connections;
• the most recent ShutdownEventArgs on those connections (used as the Description when the connection is down);
• the auto-recovery RecoverySucceededAsync event (used to track ReconnectAttempts);
• ConnectionFactory.Ssl.CertPath (parsed for NotAfter when TLS is enabled).

A reconnect within the last two minutes flips Status from Healthy to Degraded. After that window the connection is reported as Healthy again even if the reconnect counter is non-zero.

---

---

url: /guide/http/caching.md