Date: Thu, 25 Jul 2002 16:26:47 -0700 (PDT) From: Matthew Dillon <dillon@apollo.backplane.com> To: David Schultz <dschultz@uclink.Berkeley.EDU> Cc: Jaime Bozza <jbozza@thinkburst.com>, stable@FreeBSD.ORG Subject: Re: RE: Abominable NFSv3 read performance / FreeBSD server / Solaris client Message-ID: <200207252326.g6PNQl5F035612@apollo.backplane.com> References: <200207250002.g6P02m07030238@apollo.backplane.com> <02d401c233e7$49ef80d0$6401010a@bozza> <20020725183646.GA785@HAL9000.homeunix.com>
next in thread | previous in thread | raw e-mail | index | archive | help
:...
:> setting xmit_hiwat in the startup scripts and restarting the Solaris
:> server to assure the setting was changed before the nfs daemons came
:> online) I may still not be getting the settings correct, but I'm at a
:> loss at what I'm missing.
:
:Now I'm curious. What is it that makes Solaris<->Solaris
:performance good despite the TCP breakage? If the server always
:advertises a tiny window, performance ougut to be equally bad when
:talking to Solaris or FreeBSD. I've seen threads about this
:problem before on the lists, and I don't recall anyone coming up
:with a real answer.
This could also be due to FreeBSD's lowat/hiwat hysteresis. Again,
the real problem here is that the acks being returned by Solaris are
running too far behind the data that FreeBSD is sending. Solaris
appears to fall back to delayed acks after sending three acks in series,
even though it is no where near cleaning up the window, and is making
incredibly dumb assumptions as to the hysteresis on the other side
of the connection. Solaris itself obviously implements different
hysteresis points (if it implements any hysteresis at all). This really
does look like a bug in Solaris's delayed-ack implementation to me.
-Matt
Matthew Dillon
<dillon@backplane.com>
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-stable" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200207252326.g6PNQl5F035612>