HI I m trying to write a client that makes http2 request to multiple
servers(which already supports http2). so far I have written this
standalone program which works fine that I can see all requests goes via
same tcp connection(verified with the help of wireshark). I use Apache
Httpclient 5 with conscrypt to support ALPN in jdk8 (jdk8 is my
requirement, I cannot upgrade to jdk9 or later)

The Main block goes like,

try {
    TrustManager[] trustAllCerts = new TrustManager[]{
        new X509TrustManager(){
            public X509Certificate[] getAcceptedIssuers(){ return null; }
            public void checkClientTrusted(X509Certificate[] certs,
String authType) {}
            public void checkServerTrusted(X509Certificate[] certs,
String authType) {}
        }
    };
    Provider provider = Conscrypt.newProvider();

    SSLContext sslContext = SSLContext.getInstance("TLSv1.3", provider);
    sslContext.init(null, trustAllCerts, new SecureRandom());

    client = Http2AsyncClientBuilder.create().setTlsStrategy(new
ConscryptClientTlsStrategy(sslContext)).setH2Config(H2Config.DEFAULT).build();
    client.start();

    ThreadPoolExecutor tpe =
(ThreadPoolExecutor)Executors.newFixedThreadPool(50);
    for (int i=0; i < 50; i++) {
        Runnable worker = new WorkerClass(client, i);
        tpe.execute(worker);
    }} catch (Exception e) {
    e.printStackTrace();}

The runnable goes like,

static class WorkerClass implements Runnable {
    CloseableHttpAsyncClient client = null;
    int i = 0;
    WorkerClass(CloseableHttpAsyncClient client, int i) {
        this.client = client;
        this.i = i;
    }
    public void run() {
        RequestConfig requestConfig = RequestConfig.custom()
                .setConnectTimeout(Timeout.of(15,
TimeUnit.SECONDS))//Determines the timeout until a new connection is
fully established.
                .setConnectionRequestTimeout(Timeout.of(60,
TimeUnit.SECONDS))//Returns the connection lease request timeout used
when requesting a connection from the connection manager.
                .setResponseTimeout(Timeout.of(60,
TimeUnit.SECONDS))//Determines the timeout until arrival of a response
from the opposite endpoint.
                .build();

        String url =
"https://localhost:8081/myagent/getOutput?"+System.currentTimeMillis();
        String cachedXMLRequest = "<?xml version=\"1.0\"
standalone=\"no\"?><My XML REQUEST GOES HERE>";
        SimpleHttpRequest request= SimpleHttpRequests.POST.create(url);
        request.setBodyText(cachedXMLRequest, ContentType.APPLICATION_JSON);
        request.setConfig(requestConfig);
        final CountDownLatch latch = new CountDownLatch(1);
        client.execute(request, new FutureCallback<SimpleHttpResponse>() {
            @Override
            public void cancelled() {
                System.out.println("Cancelled");
                latch.countDown();
            }

            @Override
            public void completed(SimpleHttpResponse arg0) {
                System.out.println("Completed "+arg0.getBodyText());
                latch.countDown();
            }

            @Override
            public void failed(Exception exception) {
                System.out.println("Failed ");
                exception.printStackTrace();
                latch.countDown();
            }
        });

        try {
            latch.await(60, TimeUnit.SECONDS);
        } catch (InterruptedException e) {
            // TODO Auto-generated catch block
            e.printStackTrace();
        }
    }}

So I simulated this standalone in my client which asynchronously calls
multiple servers for polling using the above logic. But here I can see one
tcp connection per request and the connection is closed at the end of the
request(Though the response header returns as HTTP/2.0). I use the main
block code to initialize the client at the startup of my application and
create the url, request and response objects and execute each request
simultaneously in a threaded model(Asynchronous multiplexing). My
requirement is to use one tcp connection per domain and use it for
prolonged time for handling high number of requests. What am I missing here?

Or someone hep me with the proper way fo handling http2 using httpclient 5

Kindy shed some light on it.

SO link :
https://stackoverflow.com/questions/54548110/how-to-do-http2-correctly-with-apache-httpcient-5-beta


TIA

-- 
*With Regards,*
*Santhosh Kumar J*

Reply via email to