Is thread pool size is configurable ?

[ychaydhari]

We have successfully upgraded our Jetty server from Jetty 11 to Jetty 12. Everything is working fine. I have one question. We have created a standalone server using Jetty.

httpServer = new Server(threadPool);

In this thread, the pool object we have set the max pool size to 500, can we increase it, or can we make it configurable? Is there any impact if we increase this thread pool? Can someone please give more details about this? What is the max limit?

[joakime]

In this thread, the pool object we have set the max pool size to 500, can we increase it, or can we make it configurable?

That depends on what your threadPool implementation is.

The behavior that jetty-home setup uses is like this ...

jetty.project/jetty-ee10/jetty-ee10-demos/jetty-ee10-demo-embedded/src/main/java/org/eclipse/jetty/ee10/demos/LikeJettyXml.java

Lines 74 to 78 in dd45eef

	QueuedThreadPool threadPool = new QueuedThreadPool();
	threadPool.setMaxThreads(500);
	// Server
	Server server = new Server(threadPool);

Is there any impact if we increase this thread pool?

This depends on both your hardware, and the nature of your web application.
Some web applications work fine with a small number of threads, as their workload is optimized for that.
Other web applications need many threads.
We've seen configurations as low as 6 (some embedded systems), and as high as 15,000 (high end compute device, hosting 30,000+ connections for a game)
The types of technology you use in Jetty will impact this decision too.
Example: HTTP/2 typically needs more threads, and WebSocket is longer lived than HTTP and puts a different kind of demand on the threadpool.

Can someone please give more details about this?

See the documentation:

https://jetty.org/docs/jetty/12/programming-guide/arch/threads.html

What is the max limit?

The limit is dependent on your implementation, your java version, and your hardware.
The correct setting is only knowable by monitoring your web application in various scenarios (low load, high load, bursts of traffic, what happens if a dependent system is slow/offline/overloaded, etc).
Also note that you can tweak specific areas of your webapp with things like QoSHandler and DoSHandler to further control resources (like threads) but to specific endpoints on your server.