How to Chain Asynchronous Calls and Propagate Exceptions with CompletableFuture in Spring - java

I'm very new to multi-threading and CompletableFuture, and I wanted to know if it is possible to implement the following scenario in Spring container (Spring Boot 2) and #Async.
I have several services that basically construct a larger object, and I wanted to see if I can make the granular services run asynchronously.
I have a service (ServiceC) that depends on the output of another service (ServiceA), while the third one can run independently (ServiceB). someService is needed for all the other services, so I thought I don't need to make it an asynchronous call. Below is the layout:
#Service
public class ServiceA {
#Async
public CompletableFuture<A> doServiceAWork(int id) throws AppException {
return CompletableFuture.completedFuture(getA(id));
}
}
#Service
public class ServiceB {
#Async
public CompletableFuture<B> doServiceBWork(int id) {
return CompletableFuture.completedFuture(getB(id));
}
}
#Service
public class ServiceC {
#Async
public CompletableFuture<C> doServiceCWork(A a) {
return CompletableFuture.completedFuture(getC(a));
}
}
#Service
public class ServiceOrchestration {
// Necessary autorwiring of services
public BigObj constructBigObj(AppObj obj) throw SomeAppException {
int id = someService.getId(obj);
BigObj bigObj = new BigObj();
try {
// Conceptualizing here
CompletableFuture<A> cfA = serviceA.doServiceAWork(id);
CompletableFuture<B> cfB = serviceB.doServiceBWork(id);
// Obviously C depends on A
CompletableFuture<C> cfC = serviceB.doServiceCWork(cfA.get());
CompletableFuture.allOf(cfA, cfB, cfC).join();
bigObj.setA(cfA.get());
bigObj.setB(cfB.get());
bigObj.setC(cfC.get());
} catch (AppException | ExecutionException | InterruptedException e) {
throw new SomeAppException("..", e);
}
return BigObj;
}
}
My requirements are as following:
Invoke asynchronous call to ServiceC when ServiceA is done as the returned data should be fed to ServiceC in order to proceed
If a call to ServiceA throws an exception (AppException), it needs to be propagated to the caller -- in this case ServiceOrchestration#constructBigObject(), which handles it as above. In other word, if serviceA.doServiceAWork(id) throws an exception, every other asynchronous call should be canceled but the exception should be thrown in constructBigObj in order to be caught and handled
Wait for all the asynchronous calls to be completed then construct BigObj
For the second requirement, I looked at isCompletedExceptionally() but wasn't sure where it needs to be called from, e.g. before a call to join()? How do I get the root exception then?
How does the asynchronous chaining works here between ServiceA call and ServiceC where there are dependencies for calling one, at the same time, I need values from both, e.g. A and C, to construct BigObj?

Related

Can I return an API response without waiting on other external API calls in Spring Boot?

Is this the correct way to use #Async in Spring Boot?
#Service
class someServiceImpl {
...
public someResponseDTO getUsers(int userId) {
// Do some logic
...
// Call external API with another service method from another service impl
anotherService.emailUserInTheBackground(userId);
return someResponseDTO;
}
...
}
#Service
public class AnotherService {
#Async
public void emailUserInTheBackground(int userId) {
// This might take a while...
...
}
}
Since emailUserInTheBackground() has #Async annotation and void return type, does it block the line return someResponseDTO at all?
All I wanted is to return the response to the caller without waiting because emailUserInTheBackground() takes too long to complete and isn't directly tied to the response object.
Yes that is the correct way to run a task in the background, you can mimick the thread blocking behavior by introducing a delay.
#SpringBootApplication
#EnableAsync
public class MyApplication {
public static void main(String[] arg) {
SpringApplication.run(MyApplication.class);
}
}
then you need to mark the emailUserInTheBackground method with #Async annotation.
#Service
class AnotherService {
#Async
public void emailUserInTheBackground(int userId) {
try {
TimeUnit.SECONDS.sleep(10);
System.out.println("Print from async: "+ Thread.currentThread().getName());
} catch (InterruptedException e) {
e.printStackTrace();
}
}
}
Now add one more logger after a method call, you'll see getUsers(...) call completing first in a different thread even though the emailService thread is blocked for 10 seconds.
anotherService.emailUserInTheBackground(userId);
System.out.println("Print from service: "+ Thread.currentThread().getName());
you can also use CompletableFuture to run a task in the background.
public someResponseDTO getUsers(int userId) {
// some other task
...
// Call external API with another service method from another service impl
CompletableFuture.runAsync(() -> anotherService.emailUserInTheBackground(userId))
return someResponseDTO;
}
The relevant behavior of #Async is documented in the Spring documentation:
You can provide the #Async annotation on a method so that invocation of that method occurs asynchronously. In other words, the caller returns immediately upon invocation, while the actual execution of the method occurs in a task that has been submitted to a Spring TaskExecutor.
In the case you're describing, since the emailUserInTheBackground method is annotated with #Async and Spring's asynchronous method execution capability is enabled, the emailUserInTheBackground method will return immediately, and the call will be processed in a separate thread. The someResponseDTO value will be be returned from the getUsers method while the emailUserInTheBackground method continues to be processed in the background.

How to use CompletableFuture.thenCompose() when returning entities from repositories?

I started working with CompletableFuture in Spring Boot, and I'm seeing in some places that the usual repository methods return CompletableFuture <Entity> instead of Entity.
I do not know what is happening, but when I return instances of CompletableFuture in repositories, the code runs perfectly. However when I return entities, the code does not work asynchronously and always returns null.
Here is an example:
#Service
public class AsyncServiceImpl{
/** .. Init repository instances .. **/
#Async(AsyncConfiguration.TASK_EXECUTOR_SERVICE)
public CompletableFuture<Token> getTokenByUser(Credential credential) {
return userRepository.getUser(credential)
.thenCompose(s -> TokenRepository.getToken(s));
}
}
#Repository
public class UserRepository {
#Async(AsyncConfiguration.TASK_EXECUTOR_REPOSITORY)
public CompletableFuture<User> getUser(Credential credentials) {
return CompletableFuture.supplyAsync(() ->
new User(credentials.getUsername())
);
}
}
#Repository
public class TokenRepository {
#Async(AsyncConfiguration.TASK_EXECUTOR_REPOSITORY)
public CompletableFuture<Token> getToken(User user) {
return CompletableFuture.supplyAsync(() ->
new Token(user.getUserId())
);
}
}
The previous code runs perfectly but the following code doesn't run asynchronously and the result is always null.
#Service
public class AsyncServiceImpl {
/** .. Init repository instances .. **/
#Async(AsyncConfiguration.TASK_EXECUTOR_SERVICE)
public CompletableFuture<Token> requestToken(Credential credential) {
return CompletableFuture.supplyAsync(() -> userRepository.getUser(credential))
.thenCompose(s ->
CompletableFuture.supplyAsync(() -> TokenRepository.getToken(s)));
}
}
#Repository
public class UserRepository {
#Async(AsyncConfiguration.TASK_EXECUTOR_REPOSITORY)
public User getUser(Credential credentials) {
return new User(credentials.getUsername());
}
}
#Repository
public class TokenRepository {
#Async(AsyncConfiguration.TASK_EXECUTOR_SERVICE)
public Token getToken(User user) {
return new Token(user.getUserId());
}
}
Why doesn't this second code work?
As per the Spring #Async Javadoc:
the return type is constrained to either void or Future
and it is also further detailed in the reference documentation:
In the simplest case, the annotation may be applied to a void-returning method.
[…]
Even methods that return a value can be invoked asynchronously. However, such methods are required to have a Future typed return value. This still provides the benefit of asynchronous execution so that the caller can perform other tasks prior to calling get() on that Future.
In your second example, your #Async-annotated methods do not return a Future (or ListenableFuture and CompletableFuture which are also supported). However, Spring has to run your method asynchronously. It can thus only behave as if your method had a void return type, and thus it returns null.
As a side note, when you use #Async, your method will already run asynchronously, so you shouldn't use CompletableFuture.supplyAsync() inside the method. You should simply compute your result and return it, wrapped in CompletableFuture.completedFuture() if necessary. If your method is only composing futures (like your service that simply composes asynchronous repository results), then you probably don't need the #Async annotation. See also the example from the Getting Started guide.

Best practice to 'rollback' REST method calls inside method

The title might be incorrect, but I will try to explain my issue. My project is a Spring Boot project. I have services which do calls to external REST endpoints.
I have a service method which contains several method calls to other services I have. Every individual method call can be successful or not. Every method call is done to a REST endpoint and there can be issues that for example the webservice is not available or that it throws an unknown exception in rare cases. What ever happens, I need to be able to track which method calls were successful and if any one of them fails, I want to rollback to the original state as if nothing happened, see it a bit as #Transactional annotation. All REST calls are different endpoints and need to be called separately and are from an external party which I don't have influence on. Example:
public MyServiceImpl implements MyService {
#Autowired
private Process1Service;
#Autowired
private Process2Service;
#Autowired
private Process3Service;
#Autowired
private Process4Service;
public void bundledProcess() {
process1Service.createFileRESTcall();
process2Service.addFilePermissionsRESTcall();
process3Service.addFileMetadataRESTcall(); <-- might fail for example
process4Service.addFileTimestampRESTcall();
}
}
If for example process3Service.addFileMetadataRESTcall fails I want to do something like undo (in reverse order) for every step before process3:
process2Service.removeFilePermissionsRESTcall();
process1Service.deleteFileRESTcall();
I read about the Command pattern, but that seems to be used for Undo actions inside an application as a sort of history of actions performed, not inside a Spring web application. Is this correct for my use case too or should I track per method/webservice call if it was successful? Is there a best practice for doing this?
I guess however I track it, I need to know which method call failed and from there on perform my 'undo' method REST calls. Although in theory even these calls might also fail of course.
My main goal is to not have files being created (in my example) which any further processes have not been performed on. It should either be all successful or nothing. A sort of transactional.
Update1: improved pseudo implementation based on comments:
public Process1ServiceImpl implements Process1Service {
public void createFileRESTcall() throws MyException {
// Call an external REST api, pseudo code:
if (REST-call fails) {
throw new MyException("External REST api failed");
}
}
}
public class BundledProcessEvent {
private boolean createFileSuccess;
private boolean addFilePermissionsSuccess;
private boolean addFileMetadataSuccess;
private boolean addFileTimestampSuccess;
// Getters and setters
}
public MyServiceImpl implements MyService {
#Autowired
private Process1Service;
#Autowired
private Process2Service;
#Autowired
private Process3Service;
#Autowired
private Process4Service;
#Autowired
private ApplicationEventPublisher applicationEventPublisher;
#Transactional(rollbackOn = MyException.class)
public void bundledProcess() {
BundleProcessEvent bundleProcessEvent = new BundleProcessEvent();
this.applicationEventPublisher.publishEvent(bundleProcessEvent);
bundleProcessEvent.setCreateFileSuccess = bundprocess1Service.createFileRESTcall();
bundleProcessEvent.setAddFilePermissionsSuccess = process2Service.addFilePermissionsRESTcall();
bundleProcessEvent.setAddFileMetadataSuccess = process3Service.addFileMetadataRESTcall();
bundleProcessEvent.setAddFileTimestampSuccess = process4Service.addFileTimestampRESTcall();
}
#TransactionalEventListener(phase = TransactionPhase.AFTER_ROLLBACK)
public void rollback(BundleProcessEvent bundleProcessEvent) {
// If the last process event is successful, we should not
// be in this rollback method even
//if (bundleProcessEvent.isAddFileTimestampSuccess()) {
// remove timestamp
//}
if (bundleProcessEvent.isAddFileMetadataSuccess()) {
// remove metadata
}
if (bundleProcessEvent.isAddFilePermissionsSuccess()) {
// remove file permissions
}
if (bundleProcessEvent.isCreateFileSuccess()) {
// remove file
}
}
Your operation looks like a transaction, so you can use #Transactional annotation. From your code I can't really tell how you are managing HTTP response calls for each of those operations, but you should consider having your service methods to return them, and then do a rollback depending on response calls. You can create an array of methods like so, but how exactly you want your logic to be is up to you.
private Process[] restCalls = new Process[] {
new Process() { public void call() { process1Service.createFileRESTcall(); } },
new Process() { public void call() { process2Service.addFilePermissionsRESTcall(); } },
new Process() { public void call() { process3Service.addFileMetadataRESTcall(); } },
new Process() { public void call() { process4Service.addFileTimestampRESTcall(); } },
};
interface Process {
void call();
}
#Transactional(rollbackOn = Exception.class)
public void bundledProcess() {
restCalls[0].call();
... // say, see which process returned wrong response code
}
#TransactionalEventListener(phase = TransactionPhase.AFTER_ROLLBACK)
public void rollback() {
// handle rollback according to failed method index
}
Check this article. Might come in handy.
The answer to this question is quite broad. There are various ways to do distributed transactions to go through them all here. However, since you are using Java and Spring, your best bet is to use something like JTA (Java Transaction API), which enables a distributed transactions across multiple services/instances/etc.. Fortunately, Spring Boot supports JTA using either Atomikos or Bitronix. You can read the doc here.
One approach to enable distributed transactions is through a message broker such as JMS, RabbitMQ, Kafka, ActiveMQ, etc. and use a protocol like XA transactions (two-phase commit). In the case of external services that do not support distributed, one approach is to write a wrapper service that understands XA transactions to that external service.

Rollback Annotation in Java

I have service calls in my application that make remote network calls to other services as well as DB calls. Spring Boot has good support for rolling back bad transactions with #Transactional, but I wanted to know if I could define a custom rollback procedure using an annotation.
I would need to rollback the data on the other services as well as the database.
In code, I could do it like this:
#Transactional
public void doSomethingComplicated() {
try {
srvcOne.makeRemoteNetworkCall();
srvcTwo.makeDatabaseCall();
} catch(Exception e) {
srvcOne.rollBackNetworkCall();
}
}
but I was hoping I could do something like this:
#Transactional
#MyCustomRollbackListener(handler = MyCustomRollBackHandler.class)
public void doSomethingComplicated() {
srvcOne.makeRemoteNetworkCall();
srvcTwo.makeDatabaseCall();
}
and in the handler:
public class MyCustomRollBackHandler {
public void handleRollback() {
srvcOne.rollBackNetworkCall();
}
}
I implemented a global exception listener and I am able to get the class the exception came from, but I have no way to get the method and to retrieve any annotations on it. Here is my initial attempt:
#ControllerAdvice
public class RollbackExceptionListener{
private static final Logger logger = LoggerFactory.getLogger(RollbackExceptionListener.class);
#ExceptionHandler(Exception.class)
public void lookForAnnotationClassForException(final Exception exception) {
logger.error("Exception thrown", exception);
final StackTraceElement topElement = exception.getStackTrace()[0];
final Class callingClass = topElement.getClass();
final String methodName = topElement.getMethodName();
try {
// Can't get the method with just the name, need to
// know the params as well.
final Method method = callingClass.getMethod(methodName);
// Retrieve the annotation on the method and call the handler
} catch (NoSuchMethodException e) {
e.printStackTrace();
}
}
}
Is there anyway to do something like this?
Arguments are not part of the Stacktrace. If the method is unique, i.e. not overloaded, you can probably find it with getMethods()? Something else that comes to mind, maybe you can look at Aspects to wrap the method in some handler before it is executed. Can be done either at compile time or runtime.
The aspect can do the rollback itself, it can enrich the exception with the information you need, or it can set some ThreadLocal variable with the handler class that was defined in the method before re-throwing the exception. You can then get this value from the ThreadLocal at the point where you catch the exception.

Testing an async method call

Below is a simplified setup of my application. It has a class Foobar which calls on a facade method for fetching data. The facade then calls on a web service to actually get the data and then manipulates the data a bit and then returns it to Foobar.
Now because the web service might take a good while to run, the method call to the facade needs to be asynchronous. Hence the facade's method doesn't have a return value, but instead, the method uses a callback object. Look at the example and continue reading below.
public class Foobar {
private List<DTO> dtos;
#Autowired
private Facade facade;
public void refresh() {
facade.refreshFoobar(new CallBack() {
public void dataFetched(List<DTO> dtos) {
setDtos(dtos);
}
});
}
public void setDtos(List<DTO> dtos) {
this.dtos = dtos;
}
}
public class Facade {
...
public void refreshFoorbar(CallBack cb) {
// Fetch data from a web service
List<DTO> dtos = webService.getData();
// Manipulate DTOs
....
// call on the callback method
cb.dataFecthed(dtos);
}
}
I have two ways of making the facade's method asynchronous, either by creating a thread manually or by using springs #Async annotation.
public class Facade {
public void refreshFoorbar(CallBack cb) {
new Thread() {
#Override
public void run() {
....
}
}.start();
}
}
// ... OR ...
public class Facade {
#Async
public void refreshFoorbar(CallBack cb) {
....
}
}
My problem is that I now need to write an integration test for this chain of method calls. I think I need to force the async facade call to be synchronous when the integration test is ran, otherwise I won't know for sure when I can do the appropriate asserts. The only idea for making the method call synchronous is to use manually handled threads AND making the threading conditional (so, for testing purposes, I have an if clause which determines if the facade method should be ran in a separate thread or not).
However, I have a feeling that there could be a better solution to my problem, whether it be a better way of forcing the method to me synchronous, eg with spring, or by testing the multithreading on some way.
This is where I need your suggestions, how would you solve my problem? Note, I'm using junit for both unit and integration tests.
Simple solution would be to return a Future object like this,
#Async
public Future<String> refreshFoorbar(CallBack cb) {
yourHeavyLifting(); //asynchronous call
return new AsyncResult<String>("yourJobNameMaybe");
}
And in your test, take the future reference and call the get() method.
future.get(); // if its not already complete, waits for it to complete
assertTrue(yourTestCondition)
This blog post shows a sample.
When JUnit testing stuff like this, I use a testing callback with a CountDownLatch that gets counted down by the callback and await()ed by the test method.
private static class TestingCallback implements Callback {
private final CountDownLatch latch;
public TestingCallback(CountDownLatch latch) {
this.latch = latch;
}
#Override public void onEvent() {
this.latch.countDown();
}
}
#Test
public void testCallback() {
final CountDownLatch latch = new CountDownLatch(1);
classUnderTest.execute( new TestCallback(latch) );
assertTrue(latch.await(30, TimeUnit.SECONDS));
}
If the callback is invoked (asynchronously) by the code under test, the latch returns true and the test passes. If the callback doesn't get invoked, the test times out after thirty seconds and the assertion fails.

Categories

Resources