[Qemu-devel] Re: qemu-kvm guest network stalls

[William King]

Is this the wrong list to get help tracking down why the network stalls
under heavy load?

I have been on irc talking with multiple people, some believe the
problem is located in "xming> so it's 100% linux-bridge/ebtables", but I
have no way to confirm this.

I have been able to reproduce the network stall with two completely
different pieces of hardware. One is an AMD dual core, the other an
Intel i7, each have different network hardware, etc. Same issue.

On 06/26/2010 01:36 PM, William King wrote:
> I have an ubuntu lucid host and an ubuntu lucid guest. Under heavy
> network load the guest network stalls and is only able to come back
> after running "sudo /etc/init.d/networking restart" on the guest.
> 
> Here are some (hopefully) helpful details.
> ------------------------------------Host---------------------------
> 
> # uname -a
> Linux virtserv 2.6.32-22-server #36-Ubuntu SMP Thu Jun 3 20:38:33 UTC
> 2010 x86_64 GNU/Linux
> 
> # ifconfig -v
> br0       Link encap:Ethernet  HWaddr 00:e0:4d:20:1f:86
>           inet addr:192.168.100.145  Bcast:192.168.100.255
> Mask:255.255.255.0
>           inet6 addr: fe80::2e0:4dff:fe20:1f86/64 Scope:Link
>           UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
>           RX packets:1254011 errors:0 dropped:0 overruns:0 frame:0
>           TX packets:1070096 errors:0 dropped:0 overruns:0 carrier:0
>           collisions:0 txqueuelen:0
>           RX bytes:439139185 (439.1 MB)  TX bytes:836663432 (836.6 MB)
> 
> eth0      Link encap:Ethernet  HWaddr 00:e0:4d:20:1f:86
>           inet6 addr: fe80::2e0:4dff:fe20:1f86/64 Scope:Link
>           UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
>           RX packets:124712767 errors:0 dropped:0 overruns:0 frame:0
>           TX packets:91086029 errors:0 dropped:0 overruns:0 carrier:0
>           collisions:0 txqueuelen:1000
>           RX bytes:162562714983 (162.5 GB)  TX bytes:36005498601 (36.0 GB)
>           Interrupt:29 Base address:0x8000
> 
> lo        Link encap:Local Loopback
>           inet addr:127.0.0.1  Mask:255.0.0.0
>           inet6 addr: ::1/128 Scope:Host
>           UP LOOPBACK RUNNING  MTU:16436  Metric:1
>           RX packets:83577 errors:0 dropped:0 overruns:0 frame:0
>           TX packets:83577 errors:0 dropped:0 overruns:0 carrier:0
>           collisions:0 txqueuelen:0
>           RX bytes:15722576 (15.7 MB)  TX bytes:15722576 (15.7 MB)
> 
> virbr0    Link encap:Ethernet  HWaddr 5a:f3:dc:99:84:ee
>           inet addr:192.168.122.1  Bcast:192.168.122.255  Mask:255.255.255.0
>           inet6 addr: fe80::58f3:dcff:fe99:84ee/64 Scope:Link
>           UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
>           RX packets:0 errors:0 dropped:0 overruns:0 frame:0
>           TX packets:6 errors:0 dropped:0 overruns:0 carrier:0
>           collisions:0 txqueuelen:0
>           RX bytes:0 (0.0 B)  TX bytes:468 (468.0 B)
> 
> vnet0     Link encap:Ethernet  HWaddr ca:8b:d8:66:e6:f5
>           inet6 addr: fe80::c88b:d8ff:fe66:e6f5/64 Scope:Link
>           UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
>           RX packets:81138964 errors:0 dropped:0 overruns:0 frame:0
>           TX packets:111911797 errors:0 dropped:0 overruns:100 carrier:0
>           collisions:0 txqueuelen:500
>           RX bytes:31177137669 (31.1 GB)  TX bytes:147190153358 (147.1 GB)
> 
> # dpkg -l |grep qemu
> ii  kvm
> 1:84+dfsg-0ubuntu16+0.12.4+noroms+0ubuntu3~ppa1 dummy transitional
> pacakge from kvm to qemu-kvm
> ii  qemu-common
> 0.12.4+noroms-0ubuntu3~ppa1                     qemu common
> functionality (bios, documentation, etc)
> ii  qemu-kvm
> 0.12.4+noroms-0ubuntu3~ppa1                     Full virtualization on
> i386 and amd64 hardware
> 
> 
> I installed the 0.12.4 qemu-kvm package that was built for lucid and
> uploaded here:
> https://edge.launchpad.net/~ubuntu-virt/+archive/ppa/+packages
> 
> # cat /etc/network/interfaces
> # This file describes the network interfaces available on your system
> # and how to activate them. For more information, see interfaces(5).
> 
> # The loopback network interface
> auto lo
> iface lo inet loopback
> 
> # The primary network interface
> auto eth0
> iface eth0 inet dhcp
> 
> auto br0
> iface br0 inet dhcp
>         bridge_ports eth0
>         bridge_stp off
>         bridge_fd 0
>         bridge_maxwait 0
> 
> 
> 
> ------------------------------Guest-------------------------------
> uname -a
> Linux torrents 2.6.32-22-server #36-Ubuntu SMP Thu Jun 3 20:38:33 UTC
> 2010 x86_64 GNU/Linux
> 
> # ifconfig -v
> eth0      Link encap:Ethernet  HWaddr 52:54:00:1e:6e:b4
>           inet addr:192.168.100.203  Bcast:192.168.100.255
> Mask:255.255.255.0
>           inet6 addr: fe80::5054:ff:fe1e:6eb4/64 Scope:Link
>           UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
>           RX packets:39410626 errors:0 dropped:0 overruns:0 frame:0
>           TX packets:28376412 errors:0 dropped:0 overruns:0 carrier:0
>           collisions:0 txqueuelen:1000
>           RX bytes:52260862769 (52.2 GB)  TX bytes:10005355703 (10.0 GB)
> 
> lo        Link encap:Local Loopback
>           inet addr:127.0.0.1  Mask:255.0.0.0
>           inet6 addr: ::1/128 Scope:Host
>           UP LOOPBACK RUNNING  MTU:16436  Metric:1
>           RX packets:14096 errors:0 dropped:0 overruns:0 frame:0
>           TX packets:14096 errors:0 dropped:0 overruns:0 carrier:0
>           collisions:0 txqueuelen:0
>           RX bytes:67362075 (67.3 MB)  TX bytes:67362075 (67.3 MB)
> 
> # cat /etc/network/interfaces
> # This file describes the network interfaces available on your system
> # and how to activate them. For more information, see interfaces(5).
> 
> # The loopback network interface
> auto lo
> iface lo inet loopback
> 
> # The primary network interface
> auto eth0
> iface eth0 inet dhcp
> 
> 
> 
> 
> 
> Notes:
> The TX and RX amounts are different from guest to host because I have
> restarted the guest 'recently'.
> One interesting thing I found was that the command 'arp -an' returns
> nothing on the guest while it is stalled.
> 
> On 0.12.3 I was getting network stalls every 1-3 minutes with out fail.
> After upgrading to 0.12.4 I still get the stalls mostly every 1-5
> minutes but some times it lasts over an hour between stalls.
> 
> -William King