All 15 Java Networking & HTTP Client questions
More in Java Networking & HTTP Client
What is the modern java.net.http.HttpClient (introduced in Java 11), and how does it improve on the older HttpURLConnection for making HTTP requests? How would you add custom HTTP headers (like an Authorization bearer token) to an outgoing request using HttpRequest.Builder? How would you make an asynchronous, non-blocking HTTP request using HttpClient's sendAsync() method, and how does the returned CompletableFuture integrate with the rest of an asynchronous pipeline? How would you send a POST request with a JSON request body using HttpClient, and correctly parse a JSON response using a library like Jackson? What is the java.net.Socket and ServerSocket class pair, and how would you implement a basic TCP client-server communication using them at a lower level than HttpClient? How would you send URL-encoded form data (application/x-www-form-urlencoded) using HttpClient, as opposed to a JSON request body? How would you use InetAddress to perform DNS resolution (converting a hostname to an IP address) programmatically, and what's the difference between InetAddress.getByName() and getAllByName()? How does HttpClient handle automatic following of HTTP redirects (3xx responses), and how would you configure or disable this behavior? How would you implement a custom BodySubscriber or BodyHandler to stream a very large HTTP response body incrementally, rather than buffering the entire response into memory at once? How would you implement retry logic with exponential backoff for HTTP requests made via HttpClient, given the client doesn't provide built-in retry support natively?
Advertisement
Java › Java Networking & HTTP Client

How would you configure HttpClient with custom timeouts, a proxy, and connection pooling behavior, and what's the difference between a connection timeout and a request/response timeout?

Intermediate

Answer

HttpClient.newBuilder() lets you configure a connectTimeout (maximum time to establish the initial TCP/TLS connection) at the client level, while an individual HttpRequest.Builder's timeout() sets a per-request timeout (maximum time to wait for the complete request/response exchange) -- these are conceptually distinct: a connection timeout failure means the target server couldn't even be reached/connected to in time, while a request timeout failure means the connection succeeded but the server took too long to respond; HttpClient also supports configuring an explicit proxy (via ProxySelector) and internally manages connection pooling/reuse (particularly beneficial for HTTP/2's connection multiplexing) automatically without requiring explicit manual configuration for typical use cases.
HttpClient client = HttpClient.newBuilder()
    .connectTimeout(Duration.ofSeconds(5))   // max time to establish the connection itself
    .proxy(ProxySelector.of(new InetSocketAddress("proxy.example.com", 8080)))
    .build();

HttpRequest request = HttpRequest.newBuilder()
    .uri(URI.create("https://api.example.com/data"))
    .timeout(Duration.ofSeconds(10))          // max time for THIS specific request/response exchange
    .build();
Real-world example A service calling a historically slow, occasionally unresponsive downstream API configures both a short connectTimeout (5 seconds, since an unreachable server should fail fast) and a more generous per-request timeout (30 seconds, allowing genuinely slow-but-working responses time to complete), correctly distinguishing these two different failure modes with appropriately different tolerances.

Common follow-ups: What happens to an in-flight request if the configured timeout is exceeded -- what specific exception is thrown?;How does HttpClient's connection pooling and reuse behavior differ between HTTP/1.1 and HTTP/2 protocol versions?

HttpClient & Resilience (Polly);Diagnostics & Performance
All 15 Java Networking & HTTP Client questions