forced speed/duplex, kernel releases (Was: Linux-PowerEdge digest, Vol 1 #741 - 10 msgs)

Eric Rostetter eric.rostetter at
Thu Dec 5 20:27:01 CST 2002

Preface:  I don't mean to be disrepectful of anyone at Dell or RedHat.
I love both companies.  But until someone shows me I'm wrong, I must
assume that I'm right.  No one has yet shown me I'm wrong.  It would be
disprectful of the other folks on this list not to challange statments
I believe are wrong or misleading.

Now, I'm not saying I'm right.  I very well may be wrong.  I'm just saying,
if I'm wrong, can someone please point out why/how I'm wrong???

Thanks for indulging me...

Quoting Jeff Garzik <jgarzik at>:

> Note that Cisco has _finally_ changed their docs to indicate that
> forcing settings is not a good thing.

Because they also changed their switches to work correctly without it. :)
> However, it is also noted that (a) old Cisco docs exist :) without this
> info and (b) old Cisco switches sometimes simply do the wrong thing when
> it comes to autonegotiation.

As do some old ethernet cards.  Hence, my point was valid, no?

If I boot my cisco switch, then my machine, then it autonegotiates.  If
I boot the machine, then the switch, it fails to autonegotiate and I have no
working network. Now I can't always predict in a power outage which will boot
first.  So if I don't fix the speeds, then I don't know if my servers will
come up on the network.  If I do fix the speeds, I know they will come up.
Now how is that a bad thing?

> It is crucial to examine the effects of lack of flow control when you
> have a bunch of NICs on the network competing for the same resources.

Agreed. But it is crucial to examine the effects of your servers not being
on the network due to "broken" networking equipment also.

> > > 2) I wonder why tg3 driver isn't being used?  Folks normally report
> > > higher performance with tg3 versus bcm5700.
> >
> > Yes, it performs faster for about the first 1-3 hours until the machine
> > hangs solid, at which time the performance goes to zero.  When RedHat
> > releases a kernel with a working tg3 driver, I'll switch to it.
> this issue has been fixed.  You can download the rawhide kernel with the
> updated driver.

I'd rather wait for an official RedHat release.  AFAIK, rawhide does not
qualify.  Maybe I'm wrong.  See below.  Correct me (and your web sites)
if I'm wrong.

> > Please don't recommend using the "beta" or "unsupported" modules on a
> > production machine!
> I'm not.  Please don't be so presumptuous.

Then I must be mistaken.  I've not yet seen any official announcement, nor
anything on the RedHat errata list.  From that, it appears *to me* that there
is no supported module or kernel.

So you say it is in Rawhide.  But I thought Rawhide was your "beta" stuff.
Okay, a Google search says:

Trond Eivind Glomsrød  seawolf-list at
02 Jul 2001 11:32:33 -0400

> I keep seeing reference to Rawhide. Is this a beta version of the next
> Redhat release

It is the state of our packages the day it was pushed - it's not
tested, QAed, integrated etc.

Trond Eivind Glomsrød
Red Hat, Inc.

Okay, a year old, but doesn't sound production to me... Keep looking...

Maybe it is in this FAQ at

What is Rawhide? Where can I get development/CVS/etc from Red Hat?

Rawhide is Red Hat's public archive of development snap-shots. These are
bleeding edge, may be unstable, could be ugly, and might make your system smell
funny. They are UNSUPPORTED but feedback is appreciated via BugZilla the
almighty (and hungry). Rawhide is available at:

Hmm... Still sounds beta and unsupported...  But not too offical yet.  What
can I find that is at least semi-official.  How about the RedHat bugzilla
docs at

The problem described has already been fixed and can be obtained in the latest
version of our product.

The problem describe has been fixed in the latest development release of our
product obtainable from our ftp site.

The problem described has been fixed and will be available as an errata update
from our support web site. Please check the site to see if it is currently
available for download. 

What does that mean?  What does it say to you?  To me that says: Rawhide
is development code, not release (supported) code. Errata is code that is
supported (on the support web site). 

Is there something that says Rawhide is supported, production code?  There
may be, but I've never seen it and can't easily find it.

So, I don't know who is right.  But a search on google and a search on both give little reason to think I should trust rawhide in a
production environment.  If it is intended to be used in production code,
then that information is not getting out to RedHat users (or at least not
at my site).

> You apparently do not have the latest kernel, see above.

I've tried the latest kernels RedHat sells, and the latest kernels from
the RedHat errata page.  As far as I know, that is what I'm supposed to
do.  Maybe I'm wrong, if so, great!  But no one has ever told me that
Rawhide is supported, production code.  And I don't even remember any one
ever telling me that the fixes were even in rawhide until this message.

> 	Jeff

Again, I don't want to disrepect Jeff, RedHat, or Dell.  My goal is to
help both myself and others by furthering this discussion, and getting
the proper information out to the members of this list.

Eric Rostetter
The Department of Physics
The University of Texas at Austin

Why get even? Get odd!

More information about the Linux-PowerEdge mailing list