On 2/19/2013 8:04 PM, Jeff Bailey wrote:
On 2/19/2013 5:34 PM, Dan Kenigsberg wrote:
> On Tue, Feb 19, 2013 at 11:12:42PM +0100, Kevin Maziere Aubry wrote:
>> Hi
>>
>> I've just found a workaround ... rm
>> /usr/lib/udev/rules.d/60-net.rules and
>> reboot
>> Then I can add vlan to physical interface.
>>
>>
>>
>> 2013/2/19 Kevin Maziere Aubry <kevin.maziere(a)alterway.fr>
>>
>>> Hi
>>>
>>> Today on IRC we worked on this issue and I will try to summarized the
>>> results of our troubleshooting :
>>>
>>> Current stable systemd rpm has a bug with device mapper which cause
>>> all
>>> device created on a fibrechannel to have wrong access right.
>>> So the workaround is to replace systemd with the one on the testing
>>> repo.
>>>
>>> But the testing release as also a bug with udev which rename network
>>> interface so that each new network interface is named
>>> renameX@interface.
>>>
>>> We test some udev workaround unsucessfully. (
>>>
https://bugzilla.redhat.com/show_bug.cgi?id=907365 )
>>> Now I think a patch on systemd testing rpm should fix the issue,
>>> waiting
>>> for it
>>>
>>> Reference:
>>>
>>>
https://bugzilla.redhat.com/show_bug.cgi?id=912323
> Last word that I've heard about this bug ^^^ ("Bug 912323 - Adding a
> VLAN Device does not Work ") is that Muli can no longer reproduce it on
> his host.
>
> If this does reproduce on your system, would you provide more data as
> requested on
https://bugzilla.redhat.com/show_bug.cgi?id=912323#c2 ?
> Note that I've just called the systemd cavalry to our assistance.
I thought I had a simple (if inconvenient) work around but it looks
like something (either the deploy or syncing the management network)
puts the HWADDR line back in ifcfg-em1_1. I'm going to do some more
testing... Yep, it's the network sync that adds HWADDR back which
then triggers udev to rename em1_1.538 to em1_1 which doesn't work and
I end up with "rename??@em1_1". I can live without the sync I
suppose. I'll try leaving the manual config of the management network
alone and then config the other networks and see what happens.
Well, that didn't work. I can't seem to find any combination that plays
nicely together. Looks like Kevin's solution of ripping out udev's
ability to rename interfaces may be the only quick fix. It does get us
the ability to change ownership of LVs back which is more important. So
far, I've seen no ill effects and everything (networking and storage)
seems to be working as it should.
> Dan.
> _______________________________________________
> Users mailing list
> Users(a)ovirt.org
>
http://lists.ovirt.org/mailman/listinfo/users
_______________________________________________
Users mailing list
Users(a)ovirt.org
http://lists.ovirt.org/mailman/listinfo/users