2

我有一个用 Armeria 制作的代理服务器，它应该从源开始下载并同时将其向外流式传输到目的地。该文件应该是流式传输的，因为这是一个代理服务器，我不想让它存储和转发。这个应用程序应该处理大量的流量。

我应该在哪里寻找纯粹在 Armeria 做的任何提示？

我尝试使用 AsyncHttpClient 但我认为它正在锁定主线程或其他东西，因为请求开始累积。

下面的所有代码都在里面CompletableFuture.supplyAsync()。

有没有办法可以解决这个问题，或者更好的是，纯粹在 Armeria 中解决这个问题？

public HttpResponse doLogic(URL getUrl, String requestId, String key, Lock lock, String token) {
    // ----
    PipedOutputStream pout = new PipedOutputStream();
    BodyDeferringAsyncHandler bodyDeferringAsyncHandler = new BodyDeferringAsyncHandler(pout);
    Future<Response> respFuture = asyncHttpClient.prepareGet(getUrl.toString()).execute(bodyDeferringAsyncHandler);

    CompletableFuture<HttpResponse> r = CompletableFuture.supplyAsync(() -> {
        try (PipedInputStream pin = new PipedInputStream(pout)) {
            Response resp = bodyDeferringAsyncHandler.getResponse();
            if (resp.getStatusCode() == 200) {
                try (InputStream is = new BodyDeferringAsyncHandler.BodyDeferringInputStream(respFuture, bodyDeferringAsyncHandler, pin)) {
                    HttpHeaders headers = translateHeaders(resp.getHeaders());
                    if (headers.get("etag").contains("-")) {
                        // We remove the etag, since it comes from a multipart upload, it isn't an md5 hash of the file.
                        headers.remove("etag");
                    }

                    PipedOutputStream pout2 = new PipedOutputStream();
                    BodyDeferringAsyncHandler bodyDeferringAsyncHandler2 = new BodyDeferringAsyncHandler(pout2);
                    Future<Response> putResp = asyncHttpClient.preparePut(Config.getString("swift.proxy.base.url") + "v1/" + key)
                            .setHeaders(headers)
                            .addHeader("x-auth-token", token)
                            .setBody(is).execute(bodyDeferringAsyncHandler2);

                    Response putR = null;
                    try {
                        putR = bodyDeferringAsyncHandler2.getResponse();
                        if (putR.getStatusCode() != 201 && putR.getStatusCode() != 202) {
                            log.error("[requestId: " + requestId + "] swift returned " + putR.getStatusCode());
                            return unlockingResponse(putR.getStatusCode(), lock, key);
                        }
                        log.info("[requestId: " + requestId + "] File: " + key + " rollbacked!");
                        return unlockingResponse(200, lock, key);
                    } catch (InterruptedException e) {
                        e.printStackTrace();
                    }
                    return unlockingResponse(422, lock, key);
                }
            } else {
                log.warn("[requestId: " + requestId + "] s3 came back with status code: " + resp.getStatusCode());
                return unlockingResponse(200, lock, key);
            }
        } catch (InterruptedException | IOException e) {
            e.printStackTrace();
            return unlockingResponse(500, lock, key);
        }
    });

    return HttpResponse.from(r);
}

我希望它将文件从源流传输到目标并能够处理大量流量。

4

0 回答 0