我有一个将消息发送到远程服务的类,如下所示。 我正在使用resilience4j-retry重试网络呼叫。根据文档,由于重试实例是线程安全的,因此我将在类级别创建它并重用它。
public class RemoteMessageService {
Retry retry = Retry.of("RemoteMessageService", RetryConfig.custom()
.maxAttempts(5)
.retryExceptions(ProcessingException.class)
.intervalFunction(IntervalFunction.ofExponentialBackoff())
.build());
public void postMessageWithRetry(final String message){
Function<Integer, Void> postMessageFunction = Retry.decorateFunction(retry, this::postMessage);
try {
postMessageFunction.apply(message)
} catch (final ProcessingException e) {
LOG.warn("Got processing exception: {}", e.getMessage());
} catch (final Exception e) {
LOG.error("Got unknown exception: {}", e.getMessage());
}
}
private Void postMessage(final String message){
// Do a network call to send the message to a rest service
// throw ProcessingException in case of timeout
return null;
}
}
我的问题是Retry.decorateFunction(retry, this::postMessage);
返回的修饰函数是否也是线程安全的?
在这种情况下,我可以将其移到类级别,而不是每次调用postMessageWithRetry函数时都重复它。
答案 0 :(得分:0)
查看resilience4j-retry
代码后,我发现修饰的函数实际上是线程安全的;只要我们首先修饰的函数是线程安全的。
因此,由于postMessage
函数是线程安全的,因此我可以重写以下代码,因此经过修饰的postMessageFunction
函数也是线程安全的。
public class RemoteMessageService {
private final Retry retry = Retry.of("RemoteMessageService", RetryConfig.custom()
.maxAttempts(5)
.retryExceptions(ProcessingException.class)
.intervalFunction(IntervalFunction.ofExponentialBackoff())
.build());
private final Function<Integer, Void> postMessageFunction = Retry.decorateFunction(retry, this::postMessage);
public void postMessageWithRetry(final String message) {
try {
postMessageFunction.apply(message)
} catch (final ProcessingException e) {
LOG.warn("Got processing exception: {}", e.getMessage());
} catch (final Exception e) {
LOG.error("Got unknown exception: {}", e.getMessage());
}
}
private Void postMessage(final String message) {
// Do a network call to send the message to a rest service
// throw ProcessingException in case of timeout
return null;
}
}