[one-users] KVM Migration with 802.1q on SLES 11 SP2 results in IGMP Flood

Ruben S. Montero rsmontero at opennebula.org
Mon Oct 1 12:34:57 PDT 2012


Hi

Thanks for coming back and update...

May I ask about the other problems you experienced? Is there a way
that the drivers could be improved to help in your scenario?

Thanks for your feedback

Ruben

On Mon, Oct 1, 2012 at 4:33 PM, <Stefan.Steinert at brodos.de> wrote:
>
> Hi,
>
>
> thank you for that hint. I got it working to bring down the vlan
> interfaces but then a few other problems appeared :(
>
>
> My final solution is now to stop using the direct network management of
> opennebula.
>
>
> I now use vlan interfaces on systemside to which I have connected bridges
> and now I am able to use the default network driver without any problems.
>
>
> regards
>
>
> Stefan
>
>
>
> >>> "Ruben S. Montero" <rsmontero at opennebula.org> 8/7/2012 1:03 AM >>>
> Hi
>
>
> This could be achieved by adding the ifdown command in the clean script of
> the 802.1q driver. The core functionality is provided by HostManaged.rb. If
> you take a look at the code, maybe you can add a method similar to
>
> the activate one that iterates over the VLAN devs and execute an ip link
> set <dev> down. Note that you can only set down the interface if there are
> no other VM using it (so you would need to query the bridge). Bridge,
> devices and VLAN IDs can be easily obtained as shown in the activate method
> of the HostManaged class.
>
>
> Maybe it is easier to trace who is issuing the IGMP messages...
>
>
> Cheers
>
>
> Ruben
>
>
> On Mon, Aug 6, 2012 at 11:34 AM, <Stefan.Steinert at brodos.de> wrote:
>>
>>
>> Hello One Users,
>>
>>
>> I'm currently deploying OpenNebula on an IBM BladeCenter whose Blades OS
>> are based on SLES 11 SP2 with an attached NFS Storage for the VM Images.
>>
>> To integrate the VMs in our network I'm going to use the 802.1q Module.
>>
>>
>> The whole system is working, except one little thing :)
>>
>>
>> If I try to migrate (Live and normal) the remaining network interface on
>> the host on which the VM ran before starts to flood the whole Network with
>> IGMP queries. That results in blocking my switches (BNT Layer 2/3 cooper
>> gigabit + HP procurve 2824), so no more data will be transfered.
>>
>> If I shutdown the vlan tagged interfaces on the old machine with
>> "ifconfig eth2.100 down" for example, the network gets fully operational
>> again.
>>
>>
>> Is there already a way in Opennebula to bring down these interfaces,
>> which I only have to add the currently empty clean script?
>>
>>
>> Or does anyone know how to prevent that igmp flooding on system side?
>>
>>
>>
>> regards
>>
>>
>> Stefan
>>
>>
>> _______________________________________________
>> Users mailing list
>> Users at lists.opennebula.org
>> http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
>>
>
>
>
> --
> Ruben S. Montero, PhD
> Project co-Lead and Chief Architect
> OpenNebula - The Open Source Solution for Data Center Virtualization
> www.OpenNebula.org | rsmontero at opennebula.org | @OpenNebula




--
Ruben S. Montero, PhD
Project co-Lead and Chief Architect
OpenNebula - The Open Source Solution for Data Center Virtualization
www.OpenNebula.org | rsmontero at opennebula.org | @OpenNebula



More information about the Users mailing list