[PVE-User] Linux bridge, kvm, ipv6, intel 10gbe

[Sten Aus]

Hi again

No point making udev rule, because i40e driver is not working in my 
enivronment. Going to configure Linux bridges on 1G interface and report 
a bug for i40e driver.

Tried also older OVS (which worked for me in January). And also oldest 
i40e driver which I found at Intel's webpage.

Thanks to all and all the best!

On 28.04.15 13:55, Alexandre DERUMIER wrote:
> Hi,
>
>>> Although udev is still a problem.
> you should be able to map macaddress to specific ethX  in
>
> /etc/udev/rules.d/70-persistent-net.rules
>
> ----- Mail original -----
> De: "Sten Aus" <sten.aus at eenet.ee>
> À: "Michael Rasmussen" <mir at miras.org>, "proxmoxve" <pve-user at pve.proxmox.com>
> Envoyé: Mardi 28 Avril 2015 11:30:02
> Objet: Re: [PVE-User] Linux bridge, kvm, ipv6, intel 10gbe
>
>
>
> Hi
>
>
>
>
>
>
>
> openvswitch is the way to go.
>
>
>
>
>
>
>
>
>
> I sure hope so. :)
>
> Open vSwitch has been in my mind for some months now. I’ve tested it in January (ovs version 2.3.0-1) and got it to work, but as our cluster is quite big, I planned to go to ovs in the future.
> Anyway it seems that on that node Open vSwitch it is not working as fluently as it worked with other node in January. OVS is 2.3.1-1 and now I’ve used 10G interface. Then I used 1G interface on motherboard.
>
> Problems:
>
>      *
>
> After reboot, my 10G interface (eth5) is not connected to the bridge (udev changes interface names) and thus no packets go to network. As soon as i put eth5 to vmbr1234 everything works: ovs-vsct add-port vmbr1234 eth5 From dmesg I can see that 10G interface has names: eth1 -> eth5 and 1G appears as eth2->rename4->eth0 Despite the fact that /etc/rcS.d/ I can see that udev is 02, openvswitch 12 and networking 13th in startup row. Awkward.
>      *
>
> Also, two OVSIntPorts will not come “up” (storage88 and storage84 - maybe numbers in OVSIntPort name?).
>
>
> Okay, after some manual configuration I got OVS running, but:
>
>      *
>
> Regular/normal ping will lost 4 packets from time to time. For example, at the beginning 0-3 packets are lost somewhere, I will hear reply after 4th packet. And then same thing happens after 14th packet or so.
>      *
>
> And within ovs 2.3.0 version I was able to create OVSIntPort with “vmbr” name as well, now I need to add different name to my Internal port (within one node).
>
>
> Solutions:
>
>
>      * Downgraded i40e driver from 1.2.38 (downloaded from Intel webpage) to 1.1.23 and no OVS is almost working. Storage interfaces are still pain. Ping works correctly, but no iscsi connections are made.
>
>
> Although udev is still a problem.
>   
>
> _______________________________________________
> pve-user mailing list
> pve-user at pve.proxmox.com
> http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-user


-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 3242 bytes
Desc: S/MIME Cryptographic Signature
URL: <http://lists.proxmox.com/pipermail/pve-user/attachments/20150428/53c63548/attachment.bin>