OkHttpClient enqueue function blocked thread.

[caimo]

When I use enqueue to asyn my thread, with 2000tps, failed with large thread blocked, In spite of the dispatch thread pool of okhttpclient has been set.

check the code ,found syncronized lock, which will be the main problem.

when I use execute with CompletableFuture and custom thread pool instead of enqueue , the block problem has gone.

maybe enqueue function can be Optimized。

[MukjepScarlet]

synchoronized in enqueue was replaced with ReentrantLock at #8371

[MukjepScarlet]

Please provide test cases and more informations like Java version for it.

[caimo]

OkHttpClient 4.12.0
Jdk 17

code like this:

` okhttp3.OkHttpClient.Builder okHttpClientBuilder = new okhttp3.OkHttpClient.Builder();
okHttpClientBuilder.connectTimeout(10000, TimeUnit.MILLISECONDS);
okHttpClientBuilder.readTimeout(10000, TimeUnit.MILLISECONDS)
.writeTimeout(10000, TimeUnit.MILLISECONDS);
okHttpClientBuilder.retryOnConnectionFailure(false);
okHttpClientBuilder.connectionPool(
new ConnectionPool(10000, 60000, TimeUnit.MILLISECONDS));

    ExecutorService excutorService = new ThreadPoolExecutor(100, 200, 60L, TimeUnit.MILLISECONDS,
            new java.util.concurrent.LinkedBlockingQueue<Runnable>(10000));
    okHttpClientBuilder.dispatcher(new Dispatcher(excutorService));
    okhttp3.OkHttpClient client = okHttpClientBuilder.build();

    // just call this 2000tps at once will block thread by enqueue
    
    client.newCall(new Request("http://127.0.0.1:80/test", "GET", Headers.of(new HashMap<>())))
            .enqueue(new okhttp3.Callback() {});`

[MukjepScarlet]

100 threads parking at LinkedBlockingQueue.take

package main

import kotlinx.coroutines.CoroutineScope
import kotlinx.coroutines.Dispatchers
import kotlinx.coroutines.SupervisorJob
import kotlinx.coroutines.async
import kotlinx.coroutines.awaitAll
import kotlinx.coroutines.runBlocking
import okhttp3.*
import java.io.IOException
import java.util.*
import java.util.concurrent.ExecutorService
import java.util.concurrent.LinkedBlockingQueue
import java.util.concurrent.SynchronousQueue
import java.util.concurrent.ThreadPoolExecutor
import java.util.concurrent.TimeUnit
import kotlin.coroutines.resume
import kotlin.coroutines.resumeWithException
import kotlin.coroutines.suspendCoroutine

val scope = CoroutineScope(Dispatchers.IO + SupervisorJob())

fun main(args: Array<String>): Unit = runBlocking {
    val okHttpClientBuilder = OkHttpClient.Builder()
        .connectTimeout(10000, TimeUnit.MILLISECONDS)
        .readTimeout(10000, TimeUnit.MILLISECONDS)
        .writeTimeout(10000, TimeUnit.MILLISECONDS)
        .retryOnConnectionFailure(false)
        .connectionPool(
            ConnectionPool(10000, 60000, TimeUnit.MILLISECONDS)
        )

    val excutorService: ExecutorService = ThreadPoolExecutor(
        100, 200, 60L, TimeUnit.MILLISECONDS,
        SynchronousQueue<Runnable>() // LinkedBlockingQueue(10000)
    )
    okHttpClientBuilder.dispatcher(Dispatcher(excutorService))
    val client = okHttpClientBuilder.build()

    List(2000) {
        scope.async {
            suspendCoroutine { cont ->
                client.newCall(
                    Request.Builder().url("http://127.0.0.1:8080/").build()
                ).enqueue(object : Callback {
                    override fun onFailure(call: Call, e: IOException) {
                        cont.resumeWithException(e)
                    }
                    override fun onResponse(call: Call, response: Response) {
                        cont.resume(response)
                    }
                })
            }
        }
    }.awaitAll()
}

[MukjepScarlet]

This process stopped at 1min.
http://127.0.0.1:8080/ is Ktor CIO endpoint:

get("/") {
    call.respondText("Hello, Ktor!")
}

[caimo]

2000tps,not only 2000 requests. Need to last at least 10 minutes?
let me try it later with simple test.

[MukjepScarlet]

Well I'm curious if SpringBoot/Ktor can deal with such amount of requests... Maybe you should use WebSocket? XD

[MukjepScarlet]

Looking forward to more detailed testing, and testing with the latest beta (not yet released)