From owner-freebsd-stable Thu Jul 25 14:10:57 2002 Delivered-To: freebsd-stable@freebsd.org Received: from mx1.FreeBSD.org (mx1.FreeBSD.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 12B8037B406 for ; Thu, 25 Jul 2002 14:10:52 -0700 (PDT) Received: from mail.thinkburst.com (juno.geocomm.com [204.214.64.110]) by mx1.FreeBSD.org (Postfix) with ESMTP id 237BF43E3B for ; Thu, 25 Jul 2002 14:10:51 -0700 (PDT) (envelope-from jbozza@thinkburst.com) Received: from mailgate.thinkburstmedia.com (gateway.thinkburstmedia.com [204.214.64.100]) by mail.thinkburst.com (Postfix) with ESMTP id B327CAEA3 for ; Thu, 25 Jul 2002 16:10:51 -0500 (CDT) Received: (qmail 21908 invoked from network); 25 Jul 2002 21:10:49 -0000 From: "Jaime Bozza" To: "'Barney Wolff'" , "'Matthew Dillon'" , "'Olaf R'" Cc: Subject: RE: Abominable NFSv3 read performance / FreeBSD server / Solaris client Date: Thu, 25 Jul 2002 16:09:29 -0500 Message-ID: <027d01c2341f$908b1620$6401010a@bozza> MIME-Version: 1.0 Content-Type: text/plain; charset="US-ASCII" Content-Transfer-Encoding: 7bit X-Priority: 3 (Normal) X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook, Build 10.0.3416 Importance: Normal In-Reply-To: <027b01c2341b$29794550$6401010a@bozza> X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4910.0300 Sender: owner-freebsd-stable@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.ORG Ok, I've now been (almost) successful in getting Solaris to FreeBSD to work the same way as Solaris to Solaris. Solaris has a parameter tcp_deferred_ack_interval that is set to a default of 100. This sounded like is was causing the 1/10 second delays. I dropped it down to 10. Suddenly, NFS was tons faster at the default rsize. (I started out at 150 seconds for a 64MB file or so, and this single change moved it up to 20 seconds) I see that FreeBSD (net.inet.tcp.delacktime?) uses 100 as well. Searching Google, I came up with a few discussions on the deferred_ack parameter and in almost all cases people were lowering it to 10. The question still is why do the delays happen across platforms and not when connecting on the same platform? Jaime -----Original Message----- From: owner-freebsd-stable@FreeBSD.ORG [mailto:owner-freebsd-stable@FreeBSD.ORG] On Behalf Of Jaime Bozza Sent: Thursday, July 25, 2002 3:38 PM To: 'Barney Wolff' Cc: stable@FreeBSD.ORG Subject: RE: RE: Abominable NFSv3 read performance / FreeBSD server / Solaris client Ok, that makes sense. What doesn't make sense is that Solaris doesn't seem to handle rfc1323 all that well (my FreeBSD-client dumps showed a sliding window and that never happened with Solaris) and also never would advertise anything higher than 32K unless the FreeBSD server advertised something smaller. Almost as if they can't both be large. I don't see any setting that would explain why it does this. Either way, speed with a TCP NFS mount is horrible connecting to a FreeBSD server unless I lower rsize. I remember Matt seeing the 1/10 second delays. Perhaps there's a setting on Solaris that could fix those? I'm still going to fiddle with it from time to time, but for now I have to work on other things. Jaime To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-stable" in the body of the message