[SM-Discuss] netconf MAC setting change

Thomas Orgis thomas-forum at orgis.org
Wed Jun 21 10:50:28 EDT 2006


Am Wed, 21 Jun 2006 15:23:07 +0200
schrieb Pieter Lenaerts <e-type at sourcemage.org>: 

> On Wed, Jun 21, 2006 at 03:01:54PM +0200, Thomas Orgis wrote:
> > Am Wed, 21 Jun 2006 14:23:24 +0200
> > schrieb Arwed von Merkatz <v.merkatz at gmx.net>:
> > 
> > > On Wed, Jun 21, 2006 at 02:10:13PM +0200, Pieter Lenaerts wrote:
> > > > On Wed, Jun 21, 2006 at 12:59:24PM +0200, Thomas Orgis wrote:
> > > > > Am Tue, 20 Jun 2006 11:24:28 +0200
> > > > > schrieb Pieter Lenaerts <e-type at sourcemage.org>:
> > > > you _can_ _not_ use a volatile field as an identifier.
> > At least in theory, network scripts can keep track of their
> > renamings. They can store the "real" MAC behind the "now" MAC somewhere...
> 
> no, you are taking more complicated paths just to hold on to these mac
> variables where we could just use /sys/class/net/ fields that _never_
> change for a device on a bus in the case of name and mac changes
I agree that, though possible, remembering the MAC history may be tedious and unnecessary complex. Jut wondering now how far your "_never_ change" goes when one does changes to the hardware (moving cards in slots) that are not expected / should not break the network config.
I'm just clinging to the MAC because it was created for the purpose of identifying NICs.
I won't oppose netconf using a more safe may to identify cards internally, as long as it doesn't sacrifice flexibility that one has now.
Even when it is as simple as not _forcing_ network card identification so that the common setup of just one NIC just works and continues to work with unchanged config when nic is exchanged.
And of course my favourite example of broken config because of changed pci slot positions - I hate it when windows finds a _new_ device just because I had to reorder the cards.
As an analogy, I hate it when inserting a device into scsi bus can make sdb be called sdc instead (what is not overly easy to overcome with udev when the root is located there).

> > Fred is now called Karl.
> > 	Who is Karl?
> > The guy formerly called Fred.
> 
> this comparison does not make sense at all.
Oh, sorry - I kindof liked it;-)

> user must be able to do whatever he wants without netconf trippin
> because it was bypassed. netconf would expect mac X formerly know as
> Y while it was allready renamed to Z...
Well, I have been told that we cannot keep user from screwing things when I said a thing like that. I aggree that netconf should be as robust as possible, but you cannot catch all ideas a user may have in bypassing it.
But this is really a question about where to draw the line...

Pieter, I aggree with your reply to Arwed - let's have stable stuff in official tree and consider carefully how the future should work.


Thomas.



More information about the SM-Discuss mailing list