[sles-beta] SLES12 beta8 x86_64 net.ifnames=0 inconstisten result on BL460c-Gen8

[Jiri Bohac]

Hi,

On Tue, Jun 10, 2014 at 01:59:01PM +0000, urs.frey at post.ch wrote:
> When using
> : net.ifnames=0, I get eth1 and eth2
> : net.ifnames=1, I get eno1 and eno2
> : no-parameter, I get eth1 and eth2

net.ifnames=0 is now the default. In addition, SLE 12 now uses the same
mechanisms to provide persistent interface names as SLE 11.

That's why net.ifnames=0 behaves the same as no command-line
parameter at all. net.ifnames=1 can be used to turn on the
"predictable" interface names.

> Before until Beta7 I could use the first NIC as eth0 and in my LinuxRc info file I could define eth0 with its MAC
> This has now been somehow mixed up
> When using udev.rule: "mac=b4:b5:2f:57:a5:90,name=eth0" in my linuxrc info file I get now eth2 having MAC b4:b5:2f:57:a5:90.
> So of course the installation does fail

It seems the linuxrc parameter mac=b4:b5:2f:57:a5:90,name=eth0
does not work. Probably, the udev rule that it creates in
/etc/udev/rules.d/70-persistent-net.rules 
does not match your NIC, but the udev persistent name generator
sees eth0 as being assigned so it invents another name: eth2.

Could you please report this in Bugzilla? It would be helpful if
you could configure linuxrc to drop you in a shell and then post 
the contents of  /etc/udev/rules.d/70-persistent-net.rules

My guess is that the linuxrc generated rule did not work in
previous Betas either, but there was not peristent name
generator that would see eth0 as occupied and invent another
name.

Thanks,

-- 
Jiri Bohac <jbohac at suse.cz>
SUSE Labs, SUSE CZ