Consistent device naming

CLOSE Dave Dave.Close at us.thalesgroup.com
Wed May 14 00:59:54 UTC 2014


On 05/13/2014 05:39 PM, Russell Miller wrote:

>> I want to run the same software on all of these machines and
>> having inconsistent names /between/ the machines makes that next to
>> impossible. Using the new names means that my software has to learn
>> all those different names and can't easily determine which name
>> applies to the motherboard port. Using the old names means it can't
>> predict which name will be given to which port.
>>
>> There's got to be a better answer.

> ... don't use Fedora?
>
> Just kidding.  Sort of.

Well, that option will certainly be considered. But it appears that
nearly all of the common distributions are going to systemd/udev. Not
only would I probably still have the same problem, I'd also have to
learn a whole bunch of different things.

> Assuming since you're posting here that that's not an option :)  You
> could learn to love the default udev nomenclature, or you could add
> rules to /etc/udev.d/rules (or whatever the equivalent is) that tie
> MAC addresses to specific interface names.  It would require a little
> bit of script-fu, but assuming a competently managed key or password
> infrastructure, shouldn't take more than an hour or two.

I've done that. The problem seems to be that what I want to do is to 
/exchange/ the names assigned. One of the renames works but the other 
fails, claiming the new name is already in use. In other circumstances, 
I'd use an intermediate name, orig->temp->target, but I don't know how 
to do that with udev, or even if it is possible.

> Prepare for a whole lot more issues, though.  Upgrading from 13 is a
> significant upgrade, and there's lots that will go wrong.
> Particularly on mongrel hardware like yours.

I know. It will be a "challenge".
-- 
Dave Close, Thales Avionics, Irvine California USA.
cell +1 949 394 2124, dave.close at us.thalesgroup.com

"If a cluttered desk is a sign of a cluttered mind,
of what then is an empty desk?" --Albert Einstein


More information about the users mailing list