为了保证客户端稳定性和网络容错性，提升用户体验，在客户端建立服务端IP轮询及重试机制是十分有必要的，本文为我个人经验总结，希望对你能有参考价值。
app在启动时会请求一些配置信息，其中就包括ip的路由表，将这份路由表存储到本地，至于是sp还是对象持久化抑或是其他方式，可根据实际情况自行选择。
因为项目网络层由Retrofit+Rxjava+Okhttp实现，Retrofit运行时无法改变baseUrl,即使可以通过反射的方式来改变baseUrl，也无法对已经生成的service对象起作用，而且我的项目中所有service对象都通过Dagger2注入，所以最终使用了这样一种方式。
url交给UrlManager来管理
public class UrlManager {
public static final String[] HOST_SITE = {"https://xxx/api/",xxx};
public static final String[] HOST_WEB = {"https://xxx/",xxx};
public static final String HOST_SITE_DEBUG = "https://xxx/api/";
public static final String HOST_WEB_DEBUG = "https://xxx/";
public static List<String> list;
public static Random random = new Random();
public static String getHostSite() {
if (BuildConfig.IS_DEBUG) {
return processUrl(HOST_SITE_DEBUG);
} else {
String host = getDynamicHost();
if (!TextUtils.isEmpty(host)) return host;
return HOST_SITE[random.nextInt(HOST_SITE.length)];
}
}
@Nullable
private static String getDynamicHost() {
int index = (int) SPUtils.get(NeutronApplication.getContext(), Constants.URL, 0);
if (list != null && list.size() > 0 && index < list.size())
return list.get(index);
return null;
}
public static String getHostWeb() {
if (BuildConfig.IS_DEBUG) {
return processUrl(HOST_WEB_DEBUG);
} else {
String host = getDynamicHost();
if (!TextUtils.isEmpty(host)) return host;
return HOST_WEB[random.nextInt(HOST_WEB.length)];
}
}
public static void setHosts(List<String> list) {
UrlManager.list = list;
RxHelper.setCounterAttempts(list.size());
}
public static void updateUrlIndex(int i) {
if (list != null && i >= list.size())
i = 0;
SPUtils.put(NeutronApplication.getContext(), Constants.URL, i);
}
public static void updateUrlIndex() {
int o = (int) SPUtils.get(NeutronApplication.getContext(), Constants.URL, 0);
updateUrlIndex(o + 1);
}
app启动时拉取到配置后设置UrlManager中的路由表，然后每次根据索引去路由表动态拿请求地址，那路由索引由谁来控制呢？
因为我将项目中的rxjava抽取了一层RxHelper,所以这件事就交给RxHelper来干了，可以覆盖所有的网络请求。
public class RxHelper {
private static final int COUNTER_START = 0;
private static int COUNTER_ATTEMPTS = 3;
public static void setCounterAttempts(int counterAttempts) {
COUNTER_ATTEMPTS = counterAttempts;
}
public static <T> rx.Observable.Transformer<T, T> handleResult() {
return tObservable -> tObservable
.flatMap(RxHelper::createData)
.retryWhen(observable -> observable.compose(zipWithFlatMap()))
.observeOn(AndroidSchedulers.mainThread())
.subscribeOn(Schedulers.io());
}
public static <T> rx.Observable.Transformer<T, T> handleResultWithOutRetryPolicy() {
return tObservable -> tObservable.flatMap(RxHelper::createData)
.observeOn(AndroidSchedulers.mainThread())
.subscribeOn(Schedulers.io());
}
public static <T> Observable.Transformer<T, Long> zipWithFlatMap() {
return observable ->
observable.zipWith(Observable.range(COUNTER_START, COUNTER_ATTEMPTS),
(t, repeatAttempt) -> repeatAttempt)
.flatMap(new Func1<Integer, Observable<Long>>() {
@Override
public Observable<Long> call(Integer repeatAttempt) {
UrlManager.updateUrlIndex(repeatAttempt);
return Observable.timer(repeatAttempt * 200, TimeUnit.MILLISECONDS);
}
});
}
private static <T> Observable<T> createData(final T t) {
return Observable.create(new Observable.OnSubscribe<T>() {
@Override
public void call(Subscriber<? super T> subscriber) {
try {
subscriber.onNext(t);
subscriber.onCompleted();
} catch (Exception e) {
LogUtils.logw("Rxhelper: " + e.toString());
subscriber.onError(e);
}
}
});
}
}
这样在每次请求错误时，会递增路由表索引，继续下次请求，轮询的间隔为Observable.timer(repeatAttempt * 200, TimeUnit.MILLISECONDS);
对RxJava的retryWhen不理解的同学请移步对RxJava中.repeatWhen()和.retryWhen()操作符的思考。
之前也说了，retrofit不能修改baseUrl，反射的方式也不适合我的项目，至于利用builder生成新的retrofit对象的方式更不考虑了，那我是怎么实现运行时修改请求地址的呢？别忘了okhttp是可以添加拦截器的，在OkHttpIntercepter中：
public class OkHttpInterceptor implements Interceptor {
@Override
public Response intercept(Chain chain) throws IOException {
//配置request
Request request = chain.request();
Request.Builder requestBuilder = request.newBuilder();
String url = UrlManager.getHostSite();
Uri parse = Uri.parse(url);
String host = parse.getHost();
HttpUrl httpUrl = request.url().newBuilder().host(host).build();
requestBuilder.url(httpUrl);
Response.Builder responseBuilder = chain.proceed(requestBuilder.build()).newBuilder();
Response response = responseBuilder.build();
return response;
}
拦截请求的url,修改其host，这样整个流程就ok了，http的各种错误码的处理也是可以在拦截器中统一处理的，至于其他健壮性的考虑此处就不做过多阐述了。
有同学问我，如果想处理最后一次error通知怎么办呢？可以这样做，修改过的RxHelper如下：
public class RxHelper {
private static final int COUNTER_START = 0;
private static int COUNTER_ATTEMPTS = 3;
public static void setCounterAttempts(int counterAttempts) {
COUNTER_ATTEMPTS = counterAttempts;
}
public static <T> rx.Observable.Transformer<T, T> handleResult() {
return tObservable -> tObservable
.flatMap(RxHelper::createData)
.retryWhen(error -> delayedRetry(error))
.observeOn(AndroidSchedulers.mainThread())
.subscribeOn(Schedulers.io());
}
//猫腻主要在这个方法
private static Observable<Object> delayedRetry(Observable<? extends Throwable> error) {
return error.zipWith(Observable.range(COUNTER_START, COUNTER_ATTEMPTS + 1),
(i, repeatAttempt) -> repeatAttempt)
.flatMap(o -> {
UrlManager.updateUrlIndex(o);
LogUtils.logw("repeat: " + o);
return o < COUNTER_ATTEMPTS ? Observable.timer(o * 200, TimeUnit.MILLISECONDS)
: error.flatMap(Observable::error);
});
}
public static <T> rx.Observable.Transformer<T, T> handleResultWithOutRetryPolicy() {
return tObservable -> tObservable.flatMap(RxHelper::createData)
.observeOn(AndroidSchedulers.mainThread())
.subscribeOn(Schedulers.io());
}
private static <T> Observable<T> createData(final T t) {
return Observable.create(new Observable.OnSubscribe<T>() {
@Override
public void call(Subscriber<? super T> subscriber) {
try {
subscriber.onNext(t);
subscriber.onCompleted();
} catch (Exception e) {
subscriber.onError(e);
}
}
});
}
}
