RE: [Xen-API] openvswitch problem with XCP 1.0 RC3

[Adrian Costin <adrian@xxxxxxxxxxx>]

Hi,

[root@c5 ~]# lspci | grep Eth
01:00.0 Ethernet controller: Intel Corporation 82576 Gigabit Network Connection 
(rev 01)
01:00.1 Ethernet controller: Intel Corporation 82576 Gigabit Network Connection 
(rev 01)

[root@c5 ~]# xe network-list
uuid ( RO)                : bc5b37ac-1d72-1537-0c89-8edf77f77189
          name-label ( RW): Backnet
    name-description ( RW):
              bridge ( RO): xenbr0


uuid ( RO)                : f6719218-ed20-8614-8d13-a434c3dce67d
          name-label ( RW): vlan99
    name-description ( RW):
              bridge ( RO): xapi2


uuid ( RO)                : f88a18d5-ab30-f1a3-3b2e-367707cb7015
          name-label ( RW): Guest installer network
    name-description ( RW): Network on which guests will get assigned a private 
local IP address
              bridge ( RO): xapi0


uuid ( RO)                : a1641072-3a40-1727-4910-713d50bb635a
          name-label ( RW): Frontnet
    name-description ( RW):
              bridge ( RO): xenbr1

[root@c5 ~]# xe vlan-list
uuid ( RO)            : 95d8f26e-7513-767a-aab0-b07658d83582
      tagged-PIF ( RO): cb37ac6f-8bc6-d536-5cf5-01d2139ccfca
    untagged-PIF ( RO): 4d5a18db-ff10-8175-2f27-f4e4bce7d1f4
             tag ( RO): 99

Nothing interesting in xensource.log:

[20110225T16:10:05.961Z|debug|c5|1373|Async.VM.hard_reboot R:3f5507f06dc5|xapi] 
creating VIF devices and attaching to domain
[20110225T16:10:05.961Z|debug|c5|1373|Async.VM.hard_reboot R:3f5507f06dc5|xapi] 
Trying to attach PIF: 4d5a18db-ff10-8175-2f27-f4e4bce7d1f4
[20110225T16:10:05.961Z|debug|c5|1373|Async.VM.hard_reboot 
R:3f5507f06dc5|xenops] Device.Vif.add domid=7 devid=0 mac=ba:02:17:05:f6:64 
rate=none other_config=[] 
extra_private_keys=[ref=OpaqueRef:250d5c4b-8461-06f4-ab93-894530a259c5; 
vif-uuid=d1107976-a2f4-7c42-3173-e16c7f42b75b; 
network-uuid=f6719218-ed20-8614-8d13-a434c3dce67d]
[20110225T16:10:05.962Z|debug|c5|1373|Async.VM.hard_reboot 
R:3f5507f06dc5|xenops] adding device  B0[/local/domain/0/backend/vif/7/0]  
F7[/local/domain/7/device/vif/0]  H[/xapi/7/hotplug/vif/0]
[20110225T16:10:05.967Z|debug|c5|1373|Async.VM.hard_reboot 
R:3f5507f06dc5|hotplug] Hotplug.wait_for_plug: frontend (domid=7 | kind=vif | 
devid=0); backend (domid=0 | kind=vif | devid=0)
[20110225T16:10:05.967Z|debug|c5|1373|Async.VM.hard_reboot 
R:3f5507f06dc5|xenops] watch: watching xenstore paths: [ 
/xapi/7/hotplug/vif/0/hotplug ] with timeout 1200.000000 seconds
[20110225T16:10:06.344Z|debug|c5|284 xal_listen||event] VM (domid: 7) 
device_event = HotplugChanged on 0 {""->online}
[20110225T16:10:06.345Z|debug|c5|1373|Async.VM.hard_reboot 
R:3f5507f06dc5|hotplug] Synchronised ok with hotplug script: frontend (domid=7 
| kind=vif | devid=0); backend (domid=0 | kind=vif | devid=0)
[20110225T16:10:06.346Z|debug|c5|284 xal_listen|VM (domid: 7) device_event = 
HotplugChanged on 0 {""->online} D:b131fccb784d|event] Adding Resync.vif to 
queue
[20110225T16:10:06.346Z|debug|c5|284 xal_listen|VM (domid: 7) device_event = 
HotplugChanged on 0 {""->online} D:b131fccb784d|locking_helpers] push(per-VM 
queue, HotplugChanged(vif, 0) domid: 7); queue = [ VM (domid 6) @releaseDomain; 
DevEject(5696) domid 7; HotplugChanged(vif, 0) domid: 7 ](3)

I've tried upgrading the BIOS, compiling the latest igb kernel module from 
intel.com and enabling/disabling all NIC related settings in the BIOS 
(including SR-IOV). I've also confirmed vlans work on the exact machine and 
switch configuration using CentOS. The only thing that worked for XCP was 
switching to bridge instead of openvswitch.

Best regards,
Adrian Costin


-----Original Message-----
From: deshantm@xxxxxxxxx [mailto:deshantm@xxxxxxxxx] On Behalf Of Todd Deshane
Sent: Friday, February 25, 2011 6:41 PM
To: Adrian Costin
Cc: xen-api@xxxxxxxxxxxxxxxxxxx
Subject: Re: [Xen-API] openvswitch problem with XCP 1.0 RC3

On Fri, Feb 25, 2011 at 11:31 AM, Adrian Costin <adrian@xxxxxxxxxxx> wrote:
> Hi,
>
> No.
>
> NIC 0 with Network 0 is management interface
>
> Vlan99 is created using NIC1 and is assigned to the VPS.
>

Just in case that this is a local problem to you. Can you provide NIC 
information, xapi log, and any other relevant information that could try to 
explain the differences in both the Open vSwitch case and the bridge case?

Thanks,
Todd

> Best regards,
> Adrian Costin
>
> -----Original Message-----
> From: deshantm@xxxxxxxxx [mailto:deshantm@xxxxxxxxx] On Behalf Of Todd 
> Deshane
> Sent: Friday, February 25, 2011 6:31 PM
> To: Adrian Costin
> Cc: xen-api@xxxxxxxxxxxxxxxxxxx
> Subject: Re: [Xen-API] openvswitch problem with XCP 1.0 RC3
>
> On Fri, Feb 25, 2011 at 11:23 AM, Adrian Costin <adrian@xxxxxxxxxxx> wrote:
>> Hi,
>>
>>
>>
>> When using the openvswitch backend (which is the default) on XCP 1.0
>> RC3 network traffic for VLANs doesn't reach the virtual machine.
>>
>> Switching to bridge works without problems.
>>
>>
>>
>> Can somebody test this and confirm?
>
> Are you running into trying to do VLAN on the management interface 
> (not yet supported by XenServer/XCP as far as I know)
>
> See: 
> http://xen.markmail.org/search/?q=vlan+xcp+xenserver#query:vlan%20xcp%20xenserver+page:1+mid:7aujsddinvrlrjpy+state:results
>

_______________________________________________
xen-api mailing list
xen-api@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/mailman/listinfo/xen-api