From: Steffen Nurpmeso <steffen@sdaoden.eu>
Subject: Re: Add sysctl to disable Nagle's algorithm (RFC 896 - Congestion Control)
To: Job Snijders <job@openbsd.org>
Cc: Alexander Bluhm <bluhm@openbsd.org>, Otto Moerbeek <otto@drijf.net>, tech@openbsd.org
Date: Wed, 15 May 2024 02:29:12 +0200

Job Snijders wrote in
 <ZkPuwpO-Kye6GVlb@anton.sobornost.net>:
 |I am under the impression that the intention of LINUX_TCP_CORK is that
 |when you 'reset' (zeroize?) this particular socket option, the local
 |data queue is flushed into a TCP packet on the wire. Whereas unsetting
 |TCP_NOPUSH will not trigger a flush by itself. As I understand it NOPUSH
 |and TCP_CORK have subtle differences.

Thanks.

And i note i use

        if [ -n "${SERVER}" ]; then
                ${sysctl} -w net.ipv4.tcp_autocorking=1 net.ipv4.tcp_low_latency=0
        else
                ${sysctl} -w net.ipv4.tcp_autocorking=0 net.ipv4.tcp_low_latency=1
        fi

on Linux:

  tcp_autocorking - BOOLEAN
          Enable TCP auto corking :
          When applications do consecutive small write()/sendmsg() system calls,
          we try to coalesce these small writes as much as possible, to lower
          total amount of sent packets. This is done if at least one prior
          packet for the flow is waiting in Qdisc queues or device transmit
          queue. Applications can still use TCP_CORK for optimal behavior
          when they know how/when to uncork their sockets.

          Default : 1

--steffen
|
|Der Kragenbaer,                The moon bear,
|der holt sich munter           he cheerfully and one by one
|einen nach dem anderen runter  wa.ks himself off
|(By Robert Gernhardt)