From owner-freebsd-stable Wed Dec 26 1:46:11 2001 Delivered-To: freebsd-stable@freebsd.org Received: from gscamnlm03.wr.usgs.gov (gscamnlm03.wr.usgs.gov [130.118.4.113]) by hub.freebsd.org (Postfix) with ESMTP id 874BC37B41C; Wed, 26 Dec 2001 01:45:47 -0800 (PST) To: "Keith J" Cc: bmah@FreeBSD.ORG, freebsd-qa@FreeBSD.ORG, "Murray Stokely" , "Nevermind" , owner-freebsd-stable@FreeBSD.ORG, "Peter Ong" , "Robert Watson" , stable@FreeBSD.ORG, "Tom" Subject: Re: 4.5 PRERELEASE - Call for testing MIME-Version: 1.0 X-Mailer: Lotus Notes Release 5.0.8 June 18, 2001 Message-ID: From: "Robert L Sowders" Date: Wed, 26 Dec 2001 01:45:41 -0800 X-MIMETrack: Serialize by Router on gscamnlm03/SERVER/USGS/DOI(Release 5.0.8 |June 18, 2001) at 12/26/2001 01:45:46 AM, Serialize complete at 12/26/2001 01:45:46 AM Content-Type: multipart/alternative; boundary="=_alternative 0035A0E788256B2E_=" 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 This is a multipart message in MIME format. --=_alternative 0035A0E788256B2E_= Content-Type: text/plain; charset="us-ascii" auto is not reliable ( at least for the fxp driver ). It always succeeds when I force full duplex though. "Keith J" Sent by: owner-freebsd-stable@FreeBSD.ORG 12/26/2001 12:46 AM To: "Tom" cc: "Peter Ong" , "Robert Watson" , "Nevermind" , "Murray Stokely" , , , Subject: Re: 4.5 PRERELEASE - Call for testing Well gee Tom.... if one end doesn't respond to negotiation... are you saying the smart end will force a speed or duplex that can't possibly work? Lets say I have an old 10Mbs ONLY card... are you declaring that it gets toasted by auto-negotiate every single time? It is well known that a major issue existed with some NIC's and switches a few years back because the initial standards were not explicit enough. Some in the industry took the directives to mean one thing, and others to be different enough that "auto" was not reliable. Given that many FreeBSD users collect hardware from a few years past, I applied this reasoning as a possible culprit. I also mentioned several others, but you saw fit to remove them... evidently to make your point about "todays state" of hardware. If you want to call something "bad advice" please be specific as to the aspects and why that is so... I find people that edit responses and then hold court while passing judgement with blanket and terse responses to be rather tiresome and out of character with the ideal of a free exchange of knowledge. Bottom line, I have yet to see an expensive smart switch or inexpensive switch insist on a 10/100 mode that could not be supported when one end was "locked down". If this is not the case for you, please enlighten us all with your "personal" real world experience. On the other hand, if this is an issue with the semantics of auto-detect vs auto-negotiate... well, I can't help it if industry can't be consistent in their documentation. Perhaps I am going over the presumption line here... but I don't think most home users here are people running trunking using dot1q or Cisco isl to get their Mp3's... and for the record... even if one "auto-detects" something as simple as 10/100 it still must "negotiate" the balance of parameters, given that human intervention is not required... hence it is defacto "automatic", and thus my characterization as "auto-negotiate". As far as trunking goes... auto-negotiate trunking doesn't necessarily tell one what it finally settled on, or that it even settled on something.... and is a whole other thread entirely. It is always better to be parameter specific, and to that extent I do agree with you - if one can explicitly lock down both ends so much the better, but not all inexpensive switches can.... and my experience has told me it isn't necessary. Keith ----- Original Message ----- From: "Tom" To: "Keith J" Cc: "Peter Ong" ; "Robert Watson" ; "Nevermind" ; "Murray Stokely" ; ; ; Sent: Tuesday, December 25, 2001 1:50 PM Subject: Re: 4.5 PRERELEASE - Call for testing > > On Tue, 25 Dec 2001, Keith J wrote: > > ... > > Another possibility is you do not have a Samba problem, but a network > > problem with auto-negotiate. Some older 10/100 cards and switch combinations > > step all over each other. You should lock down one end to a specific speed > ... > > That is bad advice. Auto-negotiation is not auto-detect. If you > disable auto-negotiate at one end and specify manual settings, you must > disable it on the other end too, and specify the SAME manual settings. > Either use auto-negotiate everywhere, or use manual everywhere. I would > recommend using auto everywhere. > > Tom > > > To Unsubscribe: send mail to majordomo@FreeBSD.org > with "unsubscribe freebsd-stable" in the body of the message To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-stable" in the body of the message --=_alternative 0035A0E788256B2E_= Content-Type: text/html; charset="us-ascii"
auto is not reliable ( at least for the fxp driver ).  It always succeeds when I force full duplex though.


"Keith J" <kjohnso8@columbus.rr.com>
Sent by: owner-freebsd-stable@FreeBSD.ORG

12/26/2001 12:46 AM

       
        To:        "Tom" <tom@uniserve.com>
        cc:        "Peter Ong" <peter@haloflightleader.net>, "Robert Watson" <rwatson@FreeBSD.ORG>, "Nevermind" <never@nevermind.kiev.ua>, "Murray Stokely" <murray@FreeBSD.ORG>, <stable@FreeBSD.ORG>, <freebsd-qa@FreeBSD.ORG>, <bmah@FreeBSD.ORG>
        Subject:        Re: 4.5 PRERELEASE - Call for testing


Well gee Tom.... if one end doesn't respond to negotiation... are you saying
the smart end will force a speed or duplex that can't possibly work? Lets
say I have an old 10Mbs ONLY card... are you declaring that it gets toasted
by auto-negotiate every single time?

It is well known that a major issue existed with some NIC's and switches a
few years back because the initial standards were not explicit enough. Some
in the industry took the directives to mean one thing, and others to be
different enough that "auto" was not reliable. Given that many FreeBSD users
collect hardware from a few years past, I applied this reasoning as a
possible culprit. I also mentioned several others, but you saw fit to remove
them... evidently to make your point about "todays state" of hardware.

If you want to call something "bad advice" please be specific as to the
aspects and why that is so... I find people that edit responses and then
hold court while passing judgement with blanket and terse responses to be
rather tiresome and out of character with the ideal of a free exchange of
knowledge. Bottom line, I have yet to see an expensive smart switch or
inexpensive switch insist on a 10/100 mode that could not be supported when
one end was "locked down". If this is not the case for you, please enlighten
us all with your "personal" real world experience.

On the other hand, if this is an issue with the semantics of auto-detect vs
auto-negotiate... well, I can't help it if industry can't be consistent in
their documentation. Perhaps I am going over the presumption line here...
but I don't think most home users here are people running trunking using
dot1q or Cisco isl to get their Mp3's... and for the record... even if one
"auto-detects" something as simple as 10/100 it still must "negotiate" the
balance of parameters, given that human intervention is not required...
hence it is defacto "automatic", and thus my characterization as
"auto-negotiate".

As far as trunking goes... auto-negotiate trunking doesn't necessarily tell
one what it finally settled on, or that it even settled on something.... and
is a whole other thread entirely.

It is always better to be parameter specific, and to that extent I do agree
with you - if one can explicitly lock down both ends so much the better, but
not all inexpensive switches can.... and my experience has told me it isn't
necessary.

Keith

----- Original Message -----
From: "Tom" <tom@uniserve.com>
To: "Keith J" <kjohnso8@columbus.rr.com>
Cc: "Peter Ong" <peter@haloflightleader.net>; "Robert Watson"
<rwatson@FreeBSD.ORG>; "Nevermind" <never@nevermind.kiev.ua>; "Murray
Stokely" <murray@FreeBSD.ORG>; <stable@FreeBSD.ORG>;
<freebsd-qa@FreeBSD.ORG>; <bmah@FreeBSD.ORG>
Sent: Tuesday, December 25, 2001 1:50 PM
Subject: Re: 4.5 PRERELEASE - Call for testing


>
> On Tue, 25 Dec 2001, Keith J wrote:
>
> ...
> > Another possibility is you do not have a Samba problem, but a network
> > problem with auto-negotiate. Some older 10/100 cards and switch
combinations
> > step all over each other. You should lock down one end to a specific
speed
> ...
>
>   That is bad advice.  Auto-negotiation is not auto-detect.  If you
> disable auto-negotiate at one end and specify manual settings, you must
> disable it on the other end too, and specify the SAME manual settings.
> Either use auto-negotiate everywhere, or use manual everywhere.  I would
> recommend using auto everywhere.
>
> Tom
>
>
> To Unsubscribe: send mail to majordomo@FreeBSD.org
> with "unsubscribe freebsd-stable" in the body of the message


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-stable" in the body of the message

--=_alternative 0035A0E788256B2E_=-- To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-stable" in the body of the message