Simple, sophisticated retries.
Recurrent is a simple, zero-dependency library for performing retries. It features:
- Flexible retry policies
- Synchronous and asynchronous retries
- CompletableFuture and Java 8 functional interface integration
- Event Listeners
- Asynchronous API integration
- Invocation Tracking
Supports Java 6+ though the documentation uses lambdas for simplicity.
Recurrent supports flexible retry policies that allow you to express when retries should be performed.
A policy can allow retries on particular failures:
RetryPolicy retryPolicy = new RetryPolicy()
.retryOn(ConnectException.class, SocketException.class);
.retryOn(failure -> failure instanceof ConnectException);And for particular results or conditions:
retryPolicy
.retryWhen(null);
.retryIf(result -> result == null); We can add a fixed delay between retries:
retryPolicy.withDelay(1, TimeUnit.SECONDS);Or a delay that backs off exponentially:
retryPolicy.withBackoff(1, 30, TimeUnit.SECONDS);We can add a max number of retries and a max retry duration:
retryPolicy
.withMaxRetries(100)
.withMaxDuration(5, TimeUnit.MINUTES);We can also specify which results, failures or conditions to abort retries on:
retryPolicy
.abortWhen(true)
.abortOn(NoRouteToHostException.class)
.abortIf(result -> result == true)And of course we can combine these things into a single policy.
Once we've defined a retry policy, we can perform a retryable synchronous invocation:
// Run with retries
Recurrent.run(() -> doSomething(), retryPolicy);
// Get with retries
Connection connection = Recurrent.get(() -> connect(), retryPolicy);Asynchronous invocations can be performed and retried on a scheduled executor and return a RecurrentFuture. When the invocation succeeds or the retry policy is exceeded, the future is completed and any listeners registered against it are called:
Recurrent.get(() -> connect(), retryPolicy, executor)
.whenSuccess(connection -> log.info("Connected to {}", connection))
.whenFailure((result, failure) -> log.error("Connection attempts failed", failure));Java 8 users can use Recurrent to retry CompletableFuture calls:
Recurrent.future(() -> CompletableFuture.supplyAsync(() -> "foo")
.thenApplyAsync(value -> value + "bar")
.thenAccept(System.out::println), retryPolicy, executor);Recurrent can be used to create retryable Java 8 functional interfaces:
Function<String, Connection> connect = address -> Recurrent.get(() -> connect(address), retryPolicy);We can retry streams:
Recurrent.run(() -> Stream.of("foo").map(value -> value + "bar"), retryPolicy);Individual Stream operations:
Stream.of("foo").map(value -> Recurrent.get(() -> value + "bar", retryPolicy));Or individual CompletableFuture stages:
CompletableFuture.supplyAsync(() -> Recurrent.get(() -> "foo", retryPolicy))
.thenApplyAsync(value -> Recurrent.get(() -> value + "bar", retryPolicy));Recurrent supports event listeners that can be notified of various events such as when retries are performed and when invocations complete:
Recurrent.get(() -> connect(), retryPolicy, new Listeners<Connection>() {
public void onRetry(Connection cxn, Throwable failure, InvocationStats stats) {
log.warn("Failure #{}. Retrying.", stats.getAttemptCount());
}
public void onComplete(Connection cxn, Throwable failure) {
if (failure != null)
log.error("Connection attempts failed", failure);
else
log.info("Connected to {}", cxn);
}
});You can also register individual event listeners:
Recurrent.get(() -> connect(), retryPolicy, new Listeners<Connection>()
.whenRetry((c, f, stats) -> log.warn("Failure #{}. Retrying.", stats.getAttemptCount()))
.whenFailure((cxn, failure) -> log.error("Connection attempts failed", failure))
.whenSuccess(cxn -> log.info("Connected to {}", cxn)));Additional listeners are available via the Listeners and AsyncListeners classes. Asynchronous completion listeners can be registered via RecurrentFuture.
Recurrent can be integrated with asynchronous code that reports completion via callbacks. The ContextualRunnable and ContextualCallable interfaces provide an Invocation reference that can be used to manually perform retries or completion:
Recurrent.get(invocation ->
service.connect().whenComplete((result, failure) -> {
if (invocation.complete(result, failure))
log.info("Connected");
else if (!invocation.retry())
log.error("Connection attempts failed", failure);
}
), retryPolicy, executor);In addition to automatically performing retries, Recurrent can be used to track invocations for you, allowing you to manually retry as needed:
Invocation invocation = new Invocation(retryPolicy);
while (!invocation.isComplete()) {
try {
doSomething();
invocation.complete()
} catch (ConnectException e) {
invocation.recordFailure(e);
}
}Invocation tracking is also useful for integrating with APIs that have their own retry mechanism:
Invocation invocation = new Invocation(retryPolicy);
// On failure
if (invocation.canRetryOn(someFailure))
service.scheduleRetry(invocation.getWaitMillis(), TimeUnit.MILLISECONDS);See the RxJava example for a more detailed implementation.
Recurrent was designed to integrate nicely with existing libraries. Here are some example integrations:
For library developers, Recurrent integrates nicely into public APIs, allowing your users to configure retry policies for different opererations. One integration approach is to subclass the RetryPolicy class, then expose that as part of your API while the rest of Recurrent remains internal. Another approach is to use something like the Maven shade plugin to relocate Recurrent into your project's package structure as desired.
JavaDocs are available here.
Recurrent is a volunteer effort. If you use it and you like it, you can help by spreading the word!
Copyright 2015-2016 Jonathan Halterman - Released under the Apache 2.0 license.