EL 4.0 and how it does Networking now.

[Rick Stevens]

Waldher, Travis R wrote:
> I've learned a few things this Memorial day.
> 
> 1) lsmod - the "used by" column isn't as signifigant as it used to be
> with the 2.4.x kernels.  This actually changed after the 2.5.48 kernel
> and up.  If lspci shows the devices, and they appear to be functioning.
> As in, the interface comes up and you can use it.  That is good enough.

Yeah.

> Apparently there's been changes to the way the lsmod feature works.  I
> requested Redhat to document those changes, including their replacement,
> OR fix the lsmod command to work the way it always has.

That's more for the kernel gang--not Red Hat--but I get your meaning.

> 2) modules.conf is now modprobe.conf.  
> 	The "below" nomenclature no longer works:
> 		"below bonding e1000 bcm5700"
> 	The "options" feature appears to work the same, but I won't know
> for 	ure until I get past #3.

Yes, for 2.6 kernels, the file is now modprobe.conf.

> 3) The bonding driver.  Apparently, someone decided that no one would
> ever want to use more than 1 interface.  So they changed it to only
> support one.  When you attempt to bring up the second bonding interface,
> you get this:
> 
> 	"bonding1 device bond1 does not seem to be present, delaying
> initialization."
> 
> There is an entry in
> /usr/src/kernels/2.6.9-5.EL-smp-i686/include/linux/if_bonding.h that
> controls this.

Yeah, I see that.  Odd.

> The snag is, I'm not 100% sure where I modify it with the 2.6 kernel.
> The best documentation I've found implies that it would work on a 2.4
> kernel.  I've got a call in to Redhat for this, but I'll be upset if
> they tell me it's a development problem and I don't have development
> support.
 >
> Even if I do find the part of the header file I need to modify... how do
> I build it?  It's built as a module, not in to the kernel.  But there is
> no bonding.spec file that I could find, as is there is no rpm source to
> build from either.  So, do I recompile the kernel and the module change
> is picked up at the same time?  I don't know yet.

Well, it's easy enough to change.  You need to install the kernel source
RPM, do your "rpmbuild -bp" stuff, move the source, copy over the
config, edit the Makefile and do ye ol' "make oldconfig".  Then edit the
/usr/src/linux/include/linux/if_bonding.h, change the size and do a
"make modules" ("make modules" forces a "make bzImage" automatically).

You can force the NEW bonding module to be loaded by doing an
"insmod /usr/src/linux/drivers/net/bonding/bonding.ko", then set up your
interfaces as you wish and give it a try.  If it works the way you wish,
then you can copy the new bonding.ko module to /lib/modules/`uname 
-r`/kernel/drivers/net/bonding/bonding.ko and be done with it.

> I could get the bonding rpm from elsewhere and build it like normal, but
> I would rather use what Redhat shipped so they can't pull a "well that
> wasn't ours so it's not our problem" should there ever be a problem in
> the future.

That is an issue, and one of my major gripes with Red Hat.  I have run
several systems with two or more bonded virtual NICs and they work fine.
Why Red Hat limits it to a single device is a bit bewildering.
----------------------------------------------------------------------
- Rick Stevens, Senior Systems Engineer     rstevens at vitalstream.com -
- VitalStream, Inc.                       http://www.vitalstream.com -
-                                                                    -
-    "Hello. My PID is Inigo Montoya.  You `kill -9'-ed my parent    -
-                     process.  Prepare to vi."                      -
----------------------------------------------------------------------