Xen users - May 2009 - OpenSuse 10.3 - Xen 3.1.0 - VIF for DomU not recording transmitted bits correctly

I would like to track DomU bandwidth usage from Dom0 but my OpenSuse
Xen install is not tracking DomU bandwidth correctly. My DomU guests
are able to access the network and therefore the networking layer
appears to be working correctly. It just doesn''t report the traffic
correctly in ifconfig.

I am currently running OpenSuse10.3 with Xen 3.1.0_15042-51.3 running
a xen kernel 2.6.22.17-0.1. The installation is rather plain with
nothing added other than apache and some apache modules.


Upon creating a DomU (Ubuntu 8.04) instance named rudy I get a vif
(also named rudy) which I can observe using ifconfig on Dom0:

rudy      Link encap:Ethernet  HWaddr FE:FF:FF:FF:FF:FF
         inet6 addr: fe80::fcff:ffff:feff:ffff/64 Scope:Link
         UP BROADCAST RUNNING NOARP  MTU:1500  Metric:1
         RX packets:0 errors:0 dropped:0 overruns:0 frame:0
         TX packets:0 errors:0 dropped:13 overruns:0 carrier:0
         collisions:0 txqueuelen:32
         RX bytes:0 (0.0 b)  TX bytes:0 (0.0 b)

The problem is that RX never changes in the ifconfig view and TX only
records an increasing amount of dropped packets. The RX bytes and TX
bytes remain zero no matter how much traffic is generated by the DomU
guest. vif0.1, eth1, peth1 all record increased traffic in ifconfig
when my DomU guest is generating network traffic.

Any ideas how I can get things working? Alternatively, suggestions on
how to track DomU bandwidth usage on OpenSuse 10.3 would also be
welcome.

j.

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

On Mon, May 4, 2009 at 6:37 AM, Joseph King <joking@shaw.ca>
wrote:
> Upon creating a DomU (Ubuntu 8.04) instance named rudy I get a vif
> (also named rudy) which I can observe using ifconfig on Dom0:
>
> rudy      Link encap:Ethernet  HWaddr FE:FF:FF:FF:FF:FF
>         inet6 addr: fe80::fcff:ffff:feff:ffff/64 Scope:Link
>         UP BROADCAST RUNNING NOARP  MTU:1500  Metric:1
>         RX packets:0 errors:0 dropped:0 overruns:0 frame:0
>         TX packets:0 errors:0 dropped:13 overruns:0 carrier:0
>         collisions:0 txqueuelen:32
>         RX bytes:0 (0.0 b)  TX bytes:0 (0.0 b)
How did you create the domU?
By default the interfaces (on Xen 3.3.1 anyway) should be named
vifxxx.y or tap xxx.y, where xxx is the domain ID and y is the vif
number on domU (0 for eth0). You COULD change the names, of course.
>
> The problem is that RX never changes in the ifconfig view and TX only
> records an increasing amount of dropped packets. The RX bytes and TX
> bytes remain zero no matter how much traffic is generated by the DomU
> guest. vif0.1, eth1, peth1 all record increased traffic in ifconfig
> when my DomU guest is generating network traffic.
Is it PV or HVM guest?
HVM guest create two interfaces : tap and vif. Tap is used when using
unmodified drivers (domU will see it as realtek NIC by default), while
the vif one is used when you use PV drivers. Perhaps you''re looking at
the wrong interface. Do a "brctl show" to get list of interfaces on
that bridge.

Regards,

Fajar

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

Excellent reply. I would never have come to that conclusion and indeed
I am using an HVM guest in this instance. Thank you very much.

I have one remaining problem, I can name the vif interface, how do I
name or otherwise associate the tap interface with my guest?

In the example below, my vif interface for the guest has the name rudy
but the tap interface has the name tap0:

rudy      Link encap:Ethernet  HWaddr FE:FF:FF:FF:FF:FF
          inet6 addr: fe80::fcff:ffff:feff:ffff/64 Scope:Link
          UP BROADCAST RUNNING NOARP  MTU:1500  Metric:1
          RX packets:0 errors:0 dropped:0 overruns:0 frame:0
          TX packets:0 errors:0 dropped:7205 overruns:0 carrier:0
          collisions:0 txqueuelen:32
          RX bytes:0 (0.0 b)  TX bytes:0 (0.0 b)

tap0      Link encap:Ethernet  HWaddr 00:FF:47:A4:EB:CA
          inet6 addr: fe80::2ff:47ff:fea4:ebca/64 Scope:Link
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:13 errors:0 dropped:0 overruns:0 frame:0
          TX packets:7211 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:500
          RX bytes:1022 (1022.0 b)  TX bytes:440154 (429.8 Kb)


I see no info in either ifconfig or brctl that would let me reconcile
tap0 with rudy in an automated fashion (so that I can get at the total
RX and TX bytes).

j.

On Sun, May 3, 2009 at 7:52 PM, Fajar A. Nugraha <fajar@fajar.net>
wrote:
> On Mon, May 4, 2009 at 6:37 AM, Joseph King <joking@shaw.ca> wrote:
>> Upon creating a DomU (Ubuntu 8.04) instance named rudy I get a vif
>> (also named rudy) which I can observe using ifconfig on Dom0:
>>
>> rudy      Link encap:Ethernet  HWaddr FE:FF:FF:FF:FF:FF
>>         inet6 addr: fe80::fcff:ffff:feff:ffff/64 Scope:Link
>>         UP BROADCAST RUNNING NOARP  MTU:1500  Metric:1
>>         RX packets:0 errors:0 dropped:0 overruns:0 frame:0
>>         TX packets:0 errors:0 dropped:13 overruns:0 carrier:0
>>         collisions:0 txqueuelen:32
>>         RX bytes:0 (0.0 b)  TX bytes:0 (0.0 b)
>
> How did you create the domU?
> By default the interfaces (on Xen 3.3.1 anyway) should be named
> vifxxx.y or tap xxx.y, where xxx is the domain ID and y is the vif
> number on domU (0 for eth0). You COULD change the names, of course.
>
>>
>> The problem is that RX never changes in the ifconfig view and TX only
>> records an increasing amount of dropped packets. The RX bytes and TX
>> bytes remain zero no matter how much traffic is generated by the DomU
>> guest. vif0.1, eth1, peth1 all record increased traffic in ifconfig
>> when my DomU guest is generating network traffic.
>
> Is it PV or HVM guest?
> HVM guest create two interfaces : tap and vif. Tap is used when using
> unmodified drivers (domU will see it as realtek NIC by default), while
> the vif one is used when you use PV drivers. Perhaps you''re
looking at
> the wrong interface. Do a "brctl show" to get list of interfaces
on
> that bridge.
>
> Regards,
>
> Fajar
>
> _______________________________________________
> Xen-users mailing list
> Xen-users@lists.xensource.com
> http://lists.xensource.com/xen-users
>
_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

On Mon, May 4, 2009 at 1:02 PM, Joseph King <joking@shaw.ca>
wrote:
> Excellent reply. I would never have come to that conclusion and indeed
> I am using an HVM guest in this instance. Thank you very much.
>
> I have one remaining problem, I can name the vif interface, how do I
> name or otherwise associate the tap interface with my guest?
I don''t know about that one, sorry.
One alternative is to use Xen 3.3.1, where taps are named using their
domain id (like tap163.0).

However since the emulated NIC (the tap one) has very limited
performance anyway, why not use PV drivers?
That would mean traffic will go through your "rudy" interface, while
boosting network performance at the same. Redhat has PV drivers
included by default on RHEL5.3. You might be able to find drivers for
opensuse by searching for packages that contain "xen-vnif.ko" or
"xen-vbd.ko".

Or better yet, if you''re not using any HVM-specific feature (like
propietary binary-only driver), why not convert it to PV? That would
give domU maximum performance.

Regards,

Fajar

_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users