In a reactive microservice I'm registering to Eureka and using a #LoadBalanced WebClient to get a response from an instance. Registering in Eureka alone works, but once I add the #LoadBalanced WebClient I get following error.
2021-05-22 14:31:14.835 INFO 1852 --- [ main] com.netflix.discovery.DiscoveryClient : Getting all instance registry info from the eureka server
2021-05-22 14:31:14.985 ERROR 1852 --- [ main] scoveryClientServiceInstanceListSupplier : Exception occurred while retrieving instances for service 127.0.0.1
org.springframework.beans.factory.BeanCurrentlyInCreationException: Error creating bean with name 'scopedTarget.eurekaClient': Requested bean is currently in creation: Is there an unresolvable circular reference?
at org.springframework.beans.factory.support.AbstractBeanFactory.doGetBean(AbstractBeanFactory.java:274) ~[spring-beans-5.3.6.jar:5.3.6]
I assume it is related to the configuration eureka.client.webclient.enabled=true.
The Application
That's my application and the crucial parts of its configuration.
application.yml
eureka:
client:
serviceUrl:
defaultZone: ${vcap.services.eureka-service.credentials.uri:http://127.0.0.1:8761}/eureka/
webclient:
enabled: true
ConsumerApplication.java
#SpringBootApplication
public class ConsumerApplication {
public Mono<ServerResponse> handler(ServerRequest request) {
return webClientBuilder()
.baseUrl("http://producer")
.build()
.get()
.retrieve()
.bodyToMono(String.class)
.onErrorResume(e -> Mono.just("Error " + e.getMessage()))
.flatMap(r -> ok().bodyValue(Map.of("Producer says", r)));
}
#Bean
#LoadBalanced
public WebClient.Builder webClientBuilder(){
return WebClient.builder();
}
#Bean
RouterFunction<ServerResponse> routes() {
return route()
.GET("", this::handler)
.build();
}
public static void main(String[] args) {
SpringApplication.run(ConsumerApplication.class, args);
}
}
Eureka WebClient disabled
If instead eureka.client.webclient.enabled=false is used, everything works perfectly fine. However, I don't think this should be the solution.
DiscoveryClientOptionalArgsConfiguration : Eureka HTTP Client uses RestTemplate.
How would I go about using a #LoadBalanced WebClient together with eureka.client.webclient.enabled=true?
Had the same problem, this is how I fixed it.
#Bean
#LoadBalanced
public WebClient.Builder lbWebClient() {
return WebClient.builder();
}
#Bean
#Primary
public WebClient.Builder webClient() {
return WebClient.builder();
}
Use #Qualifier later on to used the loadbalanced one.
Related
Discover Client Properties
spring.application.name=load-balancer-client
spring.cloud.loadbalancer.ribbon.enabled=false
eureka.instance.prefer-ip-address=true
server.port=8081
spring.cloud.loadbalancer.health-check.refetch-instances=true
spring.cloud.loadbalancer.health-check.refetch-instances-interval=2s
API App Properties
spring.application.name=service-api
management.endpoints.web.exposure.include=*
eureka.instance.prefer-ip-address=true
eureka.client.healthcheck.enabled=true
My Client Load Balancer Configuration Looks like this
public class LoadBalancerConfiguration {
#Bean
public ServiceInstanceListSupplier instanceSupplier(ConfigurableApplicationContext context) {
return ServiceInstanceListSupplier.builder()
.withDiscoveryClient()
.withHealthChecks()
.build(context);
}
}
This is my Configuration class
#Configuration
#LoadBalancerClients(defaultConfiguration = LoadBalancerConfiguration.class)
public class WebClientConfiguration {
#Bean
#LoadBalanced
public WebClient.Builder builder() {
return WebClient.builder();
}
#Bean
WebClient webClient(WebClient.Builder builder) {
return builder.build();
}
}
When I start My Load balancer Client, the rest call fails with the below Error
2021-10-12 11:50:48.962 WARN 21664 --- [ Timer-0] o.s.c.l.core.RoundRobinLoadBalancer : No servers available for service: service-api
2021-10-12 11:50:48.962 WARN 21664 --- [ Timer-0] eactorLoadBalancerExchangeFilterFunction : LoadBalancer does not contain an instance for the service service-api
2021-10-12 11:50:48.963 ERROR 21664 --- [ Timer-0] reactor.core.publisher.Operators : Operator called default onErrorDropped
I alternatively tried by enabling health check by using only using only below
spring.cloud.loadbalancer.configurations=health-check and it results in Same Error
This happens only when the health check is enabled an it works fine .withHealthChecks() Uncommented
There is no #Configuration Annotation on the custom config class and I have started the registry server, api server and load balancing client in the right order as well
Can someone please let me know what the problem is?
In RestTemplate I have a custom interceptor which will log some request response details and saves to database.
my custom Interceptor:
import org.springframework.http.HttpRequest;
import org.springframework.http.client.ClientHttpRequestExecution;
import org.springframework.http.client.ClientHttpRequestInterceptor;
#Component
public class LogServices implements ClientHttpRequestInterceptor {
#Override
public ClientHttpResponse intercept(final HttpRequest request, final byte[] body, final ClientHttpRequestExecution execution) throws IOException {
final String uri = request.getURI().toString();
final ClientHttpResponse response = execution.execute(request, body);
//log request response details and save to database
return response;
RestTemplate bean configuration in springboot:
#Bean
public RestTemplate restTemplate(final RestTemplateBuilder restTemplateBuilder) {
return restTemplateBuilder
.setConnectTimeout(Duration.ofMillis(connectTimeout))
.setReadTimeout(Duration.ofMillis(readTimeout))
.build();
Add the interceptor to restTemplate bean:
#Configuration
public class LogInterceptorConfiguration {
#Autowired
public void configureLogger(final RestTemplate restTemplate, final LogServices logServices) {
final var interceptors = restTemplate.getInterceptors();
interceptors.add(logServices);
restTemplate.setInterceptors(interceptors);
}
How can I add this interceptor to FeignClient?
In application.yml:
feign:
client:
config:
default:
connectTimeout: 5000
readTimeout: 5000
request-interceptors[0]: com.api.restclient.InterceptorOne
request-interceptors[1]: com.api.log.LogServices
InterceptorOne which adds a header to every request in feign client:
#Configuration
public class InterceptorOne implements RequestInterceptor {
#Override
public void apply(RequestTemplate requestTemplate) {
requestTemplate.header("some-header", "value");
}
But I cannot add the LogServices interceptor since it does not work due to the error cannot be cast to class feign.RequestInterceptor
My guess is that the interceptor I am trying to add is a generic interceptor and not specifically request interceptor. So I want to know how do I add a generic interceptor to FeignClient similar to RestTemplate
You can add multiple interceptors as follows
feign:
client:
config:
feignName:
connectTimeout: 5000
readTimeout: 5000
loggerLevel: full
errorDecoder: com.example.SimpleErrorDecoder
retryer: com.example.SimpleRetryer
requestInterceptors:
- com.example.InterceptorOne
- com.example.LogServices
decode404: false
encoder: com.example.SimpleEncoder
decoder: com.example.SimpleDecoder
contract: com.example.SimpleContract
I have installed an Eureka server and have registered a service called pricing-service.
The Eureka dashboard shows
UP pricing-service:4ac78ca47bdbebb5fec98345c6232af0
under status.
Now I have a completely separate Spring boot web service which calls (through a WebClient instance) the pricing-service as http://pricing-service but I get "reactor.core.Exceptions$ReactiveException: java.net.UnknownHostException: No such host is known (pricing-service)"
exception.
So the Controller can't find the pricing-service by hostname.Further, how is the controller aware of the Eureka server in order to get to pricing-service? Shouldn't there be a reference to it in the application.properties of the web service? I couldn't find anything around the web.
WebClient doesn't know anything about Eureka out of the box. You need to use #LoadBalancerClient and #LoadBalanced to wire it up through the load balancer. See the docs here:
https://spring.io/guides/gs/spring-cloud-loadbalancer/
Now I have a completely separate Spring boot web service which calls (through a WebClient instance) the pricing-service as http://pricing-service
This separate service (the WebClient Service) of yours must also register itself with Eureka Server.
By default, webclient is not aware of having to use load-balancer to make calls to other eureka instances.
Here is one of the ways to enable such a WebClient bean:
#Configuration
public class MyBeanConfig {
#Bean
WebClient webClient(LoadBalancerClient lbClient) {
return WebClient.builder()
.filter(new LoadBalancerExchangeFilterFunction(lbClient))
.build();
}
}
Then, you can use this webClient bean to make calls as:
#Component
public class YourClient {
#Autowired
WebClient webClient;
public Mono<ResponseDto> makeCall() {
return webClient
.get()
.uri("http://pricing-service/")
// <-- change your body and subscribe to result
}
Note: Initializing a Bean of WebClient can be explored further here.
I had the issue that WebClient was not working with #LoadBalanced for me when I created a bean that returned WebClient. You have to create a bean for WebClient.Builder instead of just WebClient, otherwise the #LoadBalanced annotation is not working properly.
#Configuration
public class WebClientConfig {
#Bean
#LoadBalanced
public WebClient.Builder loadBalancedWebClientBuilder() {
return WebClient.builder();
}
}
I have a question about Springs RSocketRequester. I have a rsocket server and client. Client connects to this server and requests #MessageMapping endpoint. It works as expected.
But what if I restart the server. How to do automatic reconnect to rsocket server from client? Thanks
Server:
#Controller
class RSC {
#MessageMapping("pong")
public Mono<String> pong(String m) {
return Mono.just("PONG " + m);
}
}
Client:
#Bean
public RSocketRequester rSocketRequester() {
return RSocketRequester
.builder()
.connectTcp("localhost", 7000)
.block();
}
#RestController
class RST {
#Autowired
private RSocketRequester requester;
#GetMapping(path = "/ping")
public Mono<String> ping(){
return this.requester
.route("pong")
.data("TEST")
.retrieveMono(String.class)
.doOnNext(System.out::println);
}
}
Updated for Spring Framework 5.2.6+
You could achieve it with io.rsocket.core.RSocketConnector#reconnect.
#Bean
Mono<RSocketRequester> rSocketRequester(RSocketRequester.Builder rSocketRequesterBuilder) {
return rSocketRequesterBuilder
.rsocketConnector(connector -> connector
.reconnect(Retry.fixedDelay(Integer.MAX_VALUE, Duration.ofSeconds(1))))
.connectTcp("localhost", 7000);
}
#RestController
public class RST {
#Autowired
private Mono<RSocketRequester> rSocketRequesterMono;
#GetMapping(path = "/ping")
public Mono<String> ping() {
return rSocketRequesterMono.flatMap(rSocketRequester ->
rSocketRequester.route("pong")
.data("TEST")
.retrieveMono(String.class)
.doOnNext(System.out::println));
}
}
I don't think I would create a RSocketRequester bean in an application. Unlike WebClient (which has a pool of reusable connections), the RSocket requester wraps a single RSocket, i.e. a single network connection.
I think it's best to store a Mono<RSocketRequester> and subscribe to that to get an actual requester when needed. Because you don't want to create a new connection for each call, you can cache the result. Thanks to Mono retryXYZ operators, there are many ways you can refine the reconnection behavior.
You could try something like the following:
#Service
public class RSocketPingService {
private final Mono<RSocketRequester> requesterMono;
// Spring Boot is creating an auto-configured RSocketRequester.Builder bean
public RSocketPingService(RSocketRequester.Builder builder) {
this.requesterMono = builder
.dataMimeType(MediaType.APPLICATION_CBOR)
.connectTcp("localhost", 7000).retry(5).cache();
}
public Mono<String> ping() {
return this.requesterMono.flatMap(requester -> requester.route("pong")
.data("TEST")
.retrieveMono(String.class));
}
}
the answer here https://stackoverflow.com/a/58890649/2852528 is the right one. The only thing I would like to add is that reactor.util.retry.Retry has many options for configuring the logic of your retry including even logging.
So I would slightly improve the original answer, so we'd be increasing the time between the retry till riching the max value (16 sec) and before each retry log the failure - so we could monitor the activity of the connector:
#Bean
Mono<RSocketRequester> rSocketRequester(RSocketRequester.Builder builder) {
return builder.rsocketConnector(connector -> connector.reconnect(Retry.backoff(Integer.MAX_VALUE, Duration.ofSeconds(1L))
.maxBackoff(Duration.ofSeconds(16L))
.jitter(1.0D)
.doBeforeRetry((signal) -> log.error("connection error", signal.failure()))))
.connectTcp("localhost", 7000);
}
We are attempting to use Spring Cloud Gateway to setup a microservice based architecture. Currently, we have defined a route programatically:
#ServletComponentScan
#SpringBootApplication
public class GatewayApplication {
// to be passed to and used by custom filter
#Autowired
RestTemplate restTemplate;
public static void main(String[] args) {
SpringApplication.run(GatewayApplication.class, args);
}
#Bean
public RouteLocator customRouteLocator(RouteLocatorBuilder builder) {
return builder.routes()
.route("status", r -> r
.method(HttpMethod.GET)
.and()
.path("/status")
.filters(f -> f.rewritePath("/status", "/v2/status")
.filter(new AuthorizationFilter(restTemplate).apply(new Config(""))))
.uri("http://localhost:8081/"))
.build();
}
}
The above would route an incoming request /status via GET to another endpoint. We would like to apply a custom filter, which we have implemented in AuthorizationFilter. This filter, as the name implies, is another microservice which will either allow or deny an incoming request based on credentials and permissions.
Currently, the pattern we are following, which works, is to inject a Spring RestTemplate into the gateway class above, and then to pass this RestTemplate to the constructor of the filter.
However, how can this be done if we wanted to switch to using a YAML file for defining all the routes? Presumably in both cases Spring would be constructing a new filter for each incoming request. But in the case of YAML, how can we pass something in the construtor? If this cannot be done, is there any other way to inject a RestTemplate, or any other resource into a custom Spring gateway filter?
You can register your own custom GatewayFilterFactory. This allows you to provide a custom configuration, and within that configuration, you can use SpEL to reference a bean.
For example:
#Component
public class AuthenticationGatewayFilterFactory extends AbstractGatewayFilterFactory<AuthenticationGatewayFilterFactory.Config> {
public AuthenticationGatewayFilterFactory() {
super(Config.class);
}
#Override
public GatewayFilter apply(Config config) {
// TODO: Implement
}
public static class Config {
private RestTemplate restTemplate;
// TODO: Getters + Setters
}
}
Now you can use SpEL to properly reference a RestTemplate bean:
spring:
cloud:
gateway:
routes:
- id: status
uri: http://localhost:8081/
filters:
- name: Authentication
args:
restTemplate: "#{#nameOfRestTemplateBean}"
predicates:
- Path=/status
Alternatively, you could inject a RestTemplate bean within your gateway filter. For example:
#Component
public class AuthenticationGatewayFilterFactory extends AbstractGatewayFilterFactory<AuthenticationGatewayFilterFactory.Config> {
private RestTemplate restTemplate;
public AuthenticationGatewayFilterFactory(RestTemplate restTemplate) {
super(Config.class);
this.restTemplate = restTemplate;
}
#Override
public GatewayFilter apply(Config config) {
// TODO: Implement
}
public static class Config {
// TODO: Implement
}
}
The code/configuration necessary to do the inject is less complex, but it also makes it more difficult if you ever decide to put AuthenticationGatewayFilterFactory in a separate library, as the "consumers" of this library won't have any control over which RestTemplate is being injected.