Running a Java service hitting CXone APIs. The default HttpClient config seems to be choking. Getting java.util.concurrent.TimeoutException after about 200 requests. Tried tweaking the ApiConfiguration builder but can’t find the right setter for connection pool size. The docs only show basic auth s. Is there a way to inject a custom Apache HttpClient with a larger PoolingHttpClientConnectionManager? Current code just throws without a clear stack trace.
The timeout isn’t really a pool issue. It’s the default ApiConfiguration using a basic HttpClient that doesn’t handle high concurrency well. The Java SDK actually lets you inject your own HTTP client factory. You need to bypass the default builder and pass a custom ApacheHttpClient instance configured with a PoolingHttpClientConnectionManager.
Here’s how you set it up. First, create the connection manager with a max total of 200 and 50 per route. Then build the client and wrap it in the SDK’s HttpClient interface. Finally, pass that to your PlatformClient builder.
import org.apache.http.impl.client.PoolingHttpClientConnectionManager;
import com.mypurecloud.api.v2.ApiClient;
import com.mypurecloud.api.v2.auth.OAuth2Client;
// Setup connection manager
PoolingHttpClientConnectionManager connectionManager = new PoolingHttpClientConnectionManager();
connectionManager.setMaxTotal(200);
connectionManager.setDefaultMaxPerRoute(50);
// Build Apache HttpClient
CloseableHttpClient httpClient = HttpClients.custom()
.setConnectionManager(connectionManager)
.build();
// Wrap in SDK's HttpClient interface
ApiClient apiClient = new ApiClient(new OAuth2Client("https://api.mypurecloud.com"));
apiClient.setHttpClient(new ApacheHttpClient(httpClient));
// Use this apiClient in your PlatformClient builder
PlatformClient platformClient = new PlatformClient.Builder()
.withApiClient(apiClient)
.build();
This gives you full control over the connection pool. The TimeoutException usually happens because the default client reuses connections aggressively without proper keep-alive tuning. By setting setMaxTotal higher, you allow more concurrent requests. Also, make sure you’re not creating a new PlatformClient instance for every request. That kills performance instantly. Reuse the client.
The SDK docs are sparse on this. Most people just use the default builder and get stuck. This approach works for Genesys Cloud and NICE CXone APIs since they share the same underlying Java client structure. Just ensure your OAuth token refresh logic is thread-safe if you’re sharing the client across threads. The OAuth2Client handles refresh automatically, but concurrent refreshes can cause issues if not managed.
You might also want to add a retry policy. The API sometimes returns 429s under load. A simple exponential backoff helps. Use RetryHandler from Apache HttpClient. It’s not perfect, but it keeps the service running during peak hours.
Are you actually using the Genesys Cloud Java SDK or just raw Apache HttpClient? If it’s the SDK, you’re probably fighting the default ApiConfiguration instead of just swapping the client. The SDK builder has a setHttpClient method that takes an HttpClient instance. You just need to wire up a PoolingHttpClientConnectionManager with a higher setMaxTotal and setDefaultMaxPerRoute before building the CloseableHttpClient. Here’s the pattern.
PoolingHttpClientConnectionManager cm = new PoolingHttpClientConnectionManager();
cm.setMaxTotal(200);
cm.setDefaultMaxPerRoute(50);
CloseableHttpClient httpClient = HttpClients.custom()
.setConnectionManager(cm)
.build();
ApiConfiguration config = ApiConfiguration.builder()
.setHttpClient(httpClient)
.build();
Just make sure you close the client when your service shuts down. It stops the pool exhaustion errors instantly.