Messages in this thread | | | From | Eric Dumazet <> | Date | Wed, 11 Oct 2023 20:57:18 +0200 | Subject | Re: [PATCH net-next,v2] tcp: Set pingpong threshold via sysctl |
| |
On Wed, Oct 11, 2023 at 8:49 PM Haiyang Zhang <haiyangz@microsoft.com> wrote: > > > > > -----Original Message----- > > From: Stephen Hemminger <stephen@networkplumber.org> > > Sent: Tuesday, October 10, 2023 10:16 PM > > To: Haiyang Zhang <haiyangz@microsoft.com> > > Cc: Yuchung Cheng <ycheng@google.com>; linux-hyperv@vger.kernel.org; > > netdev@vger.kernel.org; KY Srinivasan <kys@microsoft.com>; > > davem@davemloft.net; edumazet@google.com; kuba@kernel.org; > > pabeni@redhat.com; corbet@lwn.net; dsahern@kernel.org; > > ncardwell@google.com; kuniyu@amazon.com; morleyd@google.com; > > mfreemon@cloudflare.com; mubashirq@google.com; linux- > > doc@vger.kernel.org; weiwan@google.com; linux-kernel@vger.kernel.org > > Subject: Re: [PATCH net-next,v2] tcp: Set pingpong threshold via sysctl > > > > On Tue, 10 Oct 2023 22:59:49 +0000 > > Haiyang Zhang <haiyangz@microsoft.com> wrote: > > > > > > > If this an application specific optimization, it should be in a socket option > > > > > rather than system wide via sysctl. > > > > Initially I had a similar comment but later decided a sysctl could > > > > still be useful if > > > > 1) the entire host (e.g. virtual machine) is dedicated to that application > > > > 2) that application is difficult to change > > > > > > Yes, the customer actually wants a global setting. But as suggested by Neal, > > > I changed it to be per-namespace to match other TCP tunables. > > > > Like congestion control choice, it could be both a sysctl and a socket option. > > The reason is that delayed ack is already controlled by socket options. > > I see. I am updating the doc and variable location for this sysctl tunable patch > as suggested by the reviewers, and will resubmit it. > > I will also work on a separate patch for the setsockopt option. > >
I am not sure about adding a socket option, and finding room in the socket structure.
See our recent effort reshuffling fields in tcp socket for better performance (stalled at this time).
I would rather experiment first with a sysctl.
| |