Fast graceful shutdown of ThreadPerTaskExecutor (when expected WAITING Threads)

[Alan Bateman]

On 21/12/2022 14:58, Arnaud Masson wrote:
>
> > The way to interrupt sockets has always been to close them. Then blocking methods will throw an exception.
>
> I didn’t test but I was hoping that “JEP 353: Reimplement the Legacy 
> Socket API” made classic socket API interruptible (like NIO) .
>
> This would make socket API so much more consistent with other blocking 
> stuff.
>
>

JEP 353 put in the foundation to enable this and JEP 425 makes use of 
it, see the Networking section [1] in the JEP and specifically:

"To allow for interruption and cancellation, the blocking I/O methods 
defined by java.net.Socket, ServerSocket, and DatagramSocket are now 
specified to be interruptible when invoked in a virtual thread"

So in the HTTP/1.1 scenario here, where a virtual thread is blocked on 
an idle persistent connection, the shutdownNow will cause the virtual 
thread to wakeup and close the connection.

-Alan

[1] https://openjdk.org/jeps/425#Networking
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.openjdk.org/pipermail/loom-dev/attachments/20221227/28bffd22/attachment.htm>