Not all service instances are "healthchecked"

To me it looks like that only the first instance of a service is being monitored:

Looking at org.eclipse.hono.service.AbstractApplication.deployServiceVerticles():

for (final ObjectFactory<? extends AbstractServiceBase<?>> serviceFactory : serviceFactories) {

    AbstractServiceBase<?> serviceInstance = serviceFactory.getObject();

    // registered
    healthCheckServer.registerHealthCheckResources(serviceInstance);

    final Future<String> deployTracker = Future.future();
    vertx.deployVerticle(serviceInstance, deployTracker.completer());
    deploymentTracker.add(deployTracker);

    for (int i = 1; i < maxInstances; i++) { // first instance has already been deployed
        serviceInstance = serviceFactory.getObject();

        // NOT registered

        log.debug("created new instance of service: {}", serviceInstance);
        final Future<String> tracker = Future.future();
        vertx.deployVerticle(serviceInstance, tracker.completer());
        deploymentTracker.add(tracker);
    }
}

Wouldn’t it make sense to use the “instances” option from the DeploymentOptions and let vertx handle the number of instances. And then invert the way the services find the health check server by using Spring functionality (injection).

[sophokles73]

@ctron We are talking about Verticle instances running inside the same vert.x instance, right?

[sophokles73]

fixed in https://github.com/eclipse/hono/commit/af79111f808638c786634c3384a66c421cda3d22