Windows XP guest lose connectivity with bridged network

[Hector Perez]

We have installed kvm on Ubuntu 12.04 LTS Host 64 bits.

There are 4 VM guest running Endian Firewall (Community Edition), and two Ubuntu 10.04 LTS servers. Aditionally 1 VM guest running Windows XP SP3 32 bits.

Except the Endian Firewall, all VM has virtio drivers for ethernet and disc I/O.

The problem only occurs on Windows XP VM. Aleatory the ethernet of this VM is loosed, and the guest don't respond any ping, or access from any port. The other linux VM has no problem, so aparently is not the problem of Bridged network that we saw in other reported bugs.

* The impact is big on this VM because RDP freeze (because is lose connectivity).
* All VM are using bridged networking.
* ONLY Windows XP VM lose conectiviy.

Configuration
---------------

Host:
CPU : AMD FX(tm)-8120 Eight-Core Processor
Memory: 8 GB

kvm --version
QEMU emulator version 1.0 (qemu-kvm-1.0), Copyright (c) 2003-2008 Fabrice Bellard

ifconfig:
br0 Link encap:Ethernet direcciónHW 02:85:2a:0e:23:61
          Direc. inet:192.168.2.50 Difus.:192.168.2.255 Másc:255.255.255.0
          Dirección inet6: fe80::62a4:4cff:fe36:3650/64 Alcance:Enlace
          ACTIVO DIFUSIÃN FUNCIONANDO MULTICAST MTU:1500 Métrica:1
          Paquetes RX:766315 errores:0 perdidos:0 overruns:0 frame:0
          Paquetes TX:526557 errores:0 perdidos:0 overruns:0 carrier:0
          colisiones:0 long.colaTX:0
          Bytes RX:173489456 (173.4 MB) TX bytes:122487582 (122.4 MB)

br1 Link encap:Ethernet direcciónHW 00:02:2a:f1:98:4b
          Direc. inet:192.168.31.250 Difus.:192.168.31.255 Másc:255.255.255.0
          Dirección inet6: fe80::202:2aff:fef1:984b/64 Alcance:Enlace
          ACTIVO DIFUSIÃN FUNCIONANDO MULTICAST MTU:1500 Métrica:1
          Paquetes RX:20251 errores:0 perdidos:13 overruns:0 frame:0
          Paquetes TX:6 errores:0 perdidos:0 overruns:0 carrier:0
          colisiones:0 long.colaTX:0
          Bytes RX:1025272 (1.0 MB) TX bytes:468 (468.0 B)

eth0 Link encap:Ethernet direcciónHW 60:a4:4c:36:36:50
          Dirección inet6: fe80::62a4:4cff:fe36:3650/64 Alcance:Enlace
          ACTIVO DIFUSIÃN FUNCIONANDO MULTICAST MTU:1500 Métrica:1
          Paquetes RX:2567234 errores:0 perdidos:0 overruns:0 frame:0
          Paquetes TX:4425676 errores:0 perdidos:0 overruns:0 carrier:0
          colisiones:0 long.colaTX:1000
          Bytes RX:359975549 (359.9 MB) TX bytes:4920592181 (4.9 GB)

eth1 Link encap:Ethernet direcciónHW 00:02:2a:f1:98:4b
          Dirección inet6: fe80::202:2aff:fef1:984b/64 Alcance:Enlace
          ACTIVO DIFUSIÃN FUNCIONANDO MULTICAST MTU:1500 Métrica:1
          Paquetes RX:2152595 errores:0 perdidos:0 overruns:0 frame:0
          Paquetes TX:1775224 errores:0 perdidos:0 overruns:0 carrier:0
          colisiones:0 long.colaTX:1000
          Bytes RX:1993489895 (1.9 GB) TX bytes:504955085 (504.9 MB)

lo Link encap:Bucle local
          Direc. inet:127.0.0.1 Másc:255.0.0.0
          Dirección inet6: ::1/128 Alcance:Anfitrión
          ACTIVO BUCLE FUNCIONANDO MTU:16436 Métrica:1
          Paquetes RX:3209934 errores:0 perdidos:0 overruns:0 frame:0
          Paquetes TX:3209934 errores:0 perdidos:0 overruns:0 carrier:0
          colisiones:0 long.colaTX:0
          Bytes RX:509044768 (509.0 MB) TX bytes:509044768 (509.0 MB)

tap0 Link encap:Ethernet direcciónHW 2e:39:8b:0d:56:8d
          Dirección inet6: fe80::2c39:8bff:fe0d:568d/64 Alcance:Enlace
          ACTIVO DIFUSIÃN FUNCIONANDO MULTICAST MTU:1500 Métrica:1
          Paquetes RX:1512659 errores:0 perdidos:0 overruns:0 frame:0
          Paquetes TX:1520648 errores:0 perdidos:0 overruns:0 carrier:0
          colisiones:0 long.colaTX:500
          Bytes RX:649922309 (649.9 MB) TX bytes:303778199 (303.7 MB)

tap1 Link encap:Ethernet direcciónHW 42:05:70:5b:70:2c
          Dirección inet6: fe80::4005:70ff:fe5b:702c/64 Alcance:Enlace
          ACTIVO DIFUSIÃN FUNCIONANDO MULTICAST MTU:1500 Métrica:1
          Paquetes RX:7 errores:0 perdidos:0 overruns:0 frame:0
          Paquetes TX:110369 errores:0 perdidos:0 overruns:0 carrier:0
          colisiones:0 long.colaTX:500
          Bytes RX:558 (558.0 B) TX bytes:15215756 (15.2 MB)

tap2 Link encap:Ethernet direcciónHW 56:74:0c:4c:73:6b
          Dirección inet6: fe80::5474:cff:fe4c:736b/64 Alcance:Enlace
          ACTIVO DIFUSIÃN FUNCIONANDO MULTICAST MTU:1500 Métrica:1
          Paquetes RX:1775212 errores:0 perdidos:0 overruns:0 frame:0
          Paquetes TX:2139066 errores:0 perdidos:0 overruns:0 carrier:0
          colisiones:0 long.colaTX:500
          Bytes RX:504954149 (504.9 MB) TX bytes:1988787563 (1.9 GB)

tap3 Link encap:Ethernet direcciónHW 82:4f:d9:8c:fe:7a
          Dirección inet6: fe80::804f:d9ff:fe8c:fe7a/64 Alcance:Enlace
          ACTIVO DIFUSIÃN FUNCIONANDO MULTICAST MTU:1500 Métrica:1
          Paquetes RX:1894552 errores:0 perdidos:0 overruns:0 frame:0
          Paquetes TX:1998208 errores:0 perdidos:0 overruns:0 carrier:0
          colisiones:0 long.colaTX:500
          Bytes RX:855437238 (855.4 MB) TX bytes:799568209 (799.5 MB)

tap4 Link encap:Ethernet direcciónHW 46:01:af:8e:51:cc
          Dirección inet6: fe80::4401:afff:fe8e:51cc/64 Alcance:Enlace
          ACTIVO DIFUSIÃN FUNCIONANDO MULTICAST MTU:1500 Métrica:1
          Paquetes RX:2124784 errores:0 perdidos:0 overruns:0 frame:0
          Paquetes TX:1780176 errores:0 perdidos:0 overruns:0 carrier:0
          colisiones:0 long.colaTX:500
          Bytes RX:1434067154 (1.4 GB) TX bytes:263588971 (263.5 MB)

tap5 Link encap:Ethernet direcciónHW 12:32:ee:97:bf:99
          Dirección inet6: fe80::1032:eeff:fe97:bf99/64 Alcance:Enlace
          ACTIVO DIFUSIÃN FUNCIONANDO MULTICAST MTU:1500 Métrica:1
          Paquetes RX:4399585 errores:0 perdidos:0 overruns:0 frame:0
          Paquetes TX:3249451 errores:0 perdidos:0 overruns:2010 carrier:0
          colisiones:0 long.colaTX:500
          Bytes RX:4663476724 (4.6 GB) TX bytes:1519401078 (1.5 GB)

tap6 Link encap:Ethernet direcciónHW 02:85:2a:0e:23:61
          Dirección inet6: fe80::85:2aff:fe0e:2361/64 Alcance:Enlace
          ACTIVO DIFUSIÃN FUNCIONANDO MULTICAST MTU:1500 Métrica:1
          Paquetes RX:1426831 errores:0 perdidos:0 overruns:0 frame:0
          Paquetes TX:1246727 errores:0 perdidos:0 overruns:0 carrier:0
          colisiones:0 long.colaTX:500
          Bytes RX:141508663 (141.5 MB) TX bytes:315112961 (315.1 MB)

virbr0 Link encap:Ethernet direcciónHW f2:59:6c:07:25:f4
          Direc. inet:192.168.122.1 Difus.:192.168.122.255 Másc:255.255.255.0
          ACTIVO DIFUSIÃN MULTICAST MTU:1500 Métrica:1
          Paquetes RX:0 errores:0 perdidos:0 overruns:0 frame:0
          Paquetes TX:0 errores:0 perdidos:0 overruns:0 carrier:0
          colisiones:0 long.colaTX:0
          Bytes RX:0 (0.0 B) TX bytes:0 (0.0 B)

/etc/network/interfaces:
auto eth0
iface eth0 inet manual

auto eth1
iface eth1 inet manual

auto br0
iface br0 inet static
        address 192.168.2.50
        network 192.168.2.0
        netmask 255.255.255.0
        broadcast 192.168.2.255
        gateway 192.168.2.15
        dns-nameservers 192.168.2.15
        bridge_ports eth0
        bridge_fd 9
        bridge_hello 2
        bridge_maxage 12
        bridge_stp off

auto br1
iface br1 inet static
        address 192.168.31.250
        network 192.168.31.0
        netmask 255.255.255.0
        broadcast 192.168.31.255
        bridge_ports eth1
        bridge_fd 9
        bridge_hello 2
        bridge_maxage 12
        bridge_stp off

We are executing KVM VMs from command line:

/usr/bin/kvm -hda /var/cache/convirt/vm_disks/endian.disk.xm -cdrom /dev/cdrom -net nic,vlan=0,macaddr=00:16:3e:1a:6b:53 -net tap,vlan=0,script=/etc/kvm/qemu-ifup-br0 -net nic,vlan=1,macaddr=00:16:3e:1c:33:45 -net tap,vlan=1,script=/etc/kvm/qemu-ifup-br0 -net nic,vlan=2,macaddr=00:16:3e:3e:84:b6 -net tap,vlan=2,script=/etc/kvm/qemu-ifup-br1 -usb -boot c -m 512 -vnc :52 -name EFW -usbdevice tablet -uuid b1c10838-3cb2-b20c-fbf5-0cc5340b7c09 -smp 1 -monitor unix:/var/run/kvm/monitors/endian,server,nowait -pidfile /var/run/kvm/pids/endian -daemonize

/usr/bin/kvm -cdrom /dev/cdrom -net nic,vlan=0,macaddr=00:16:3e:0c:c7:70,model=virtio -net tap,vlan=0,script=/etc/kvm/qemu-ifup-br0 -usb -boot c -m 512 -vnc :50 -uuid 11ce6593-d512-c069-3ea3-07fadf29c478 -name Ubuntu10.04_ldap -usbdevice tablet -drive file=/var/cache/convirt/vm_disks/Ubuntu10.04_l.disk01.xm,index=0,if=virtio,cache=none,media=disk -drive file=/var/cache/convirt/vm_disks/Ubuntu10.04_l.disk02.xm,index=1,if=virtio,cache=none,media=disk -drive file=/var/cache/convirt/vm_disks/Ubuntu10.04_l.disk03.xm,index=2,if=virtio,cache=none,media=disk -smp 1 -k es -monitor unix:/var/run/kvm/monitors/Ubuntu10.04_l,server,nowait -pidfile /var/run/kvm/pids/Ubuntu10.04_l -daemonize

/usr/bin/kvm -cdrom /dev/cdrom -net nic,vlan=0,macaddr=00:16:3e:5d:ee:52,model=virtio -net tap,vlan=0,script=/etc/kvm/qemu-ifup-br0 -usb -boot c -m 512 -vnc :51 -uuid 6bb091c2-fc3a-9563-e428-229a6a2d8440 -name Ubuntu10.04_2 -usbdevice tablet -drive file=/var/cache/convirt/vm_disks/Ubuntu10.04_2.disk.xm,index=0,if=virtio,cache=none,media=disk -drive file=/var/cache/convirt/vm_disks/Ubuntu10.04_2.disk2.xm,index=1,if=virtio,cache=none,media=disk -smp 1 -k es -monitor unix:/var/run/kvm/monitors/Ubuntu10.04_2,server,nowait -pidfile /var/run/kvm/pids/Ubuntu10.04_2 -daemonize

/usr/bin/kvm -cdrom /dev/cdrom -net nic,vlan=0,macaddr=00:16:3e:79:30:72,model=virtio -net tap,vlan=0,script=/etc/kvm/qemu-ifup-br0 -usb -boot c -m 2048 -no-acpi -vnc :53 -uuid ba5ad121-1844-ab94-223d-867e8eebb3dd -name WindowsXP -usbdevice tablet -drive file=/var/cache/convirt/vm_disks/WindowsXP.disk.xm,if=virtio,cache=none,media=disk -smp 2 -k es -monitor unix:/var/run/kvm/monitors/WindowsXP,server,nowait -pidfile /var/run kvm/pids/WindowsXP -daemonize

/usr/bin/kvm -cdrom /dev/cdrom -net nic,vlan=0,macaddr=00:16:3e:6d:96:4c,model=virtio -net tap,vlan=0,script=/etc/kvm/qemu-ifup-br0 -usb -boot c -m 512 -vnc :54 -uuid d01d0325-d9d5-1494-2b35-e8ce281c0a85 -name Ubuntu10.04_Nagios -usbdevice tablet -drive file=/var/cache/convirt/vm_disks/Ubuntu10.04_3.disk.xm,if=virtio,cache=none,media=disk -smp 1 -k es -monitor unix:/var/run/kvm/monitors/Ubuntu10.04_3,server,nowait -pidfile /var/run/kvm/pids/Ubuntu10.04_3 -daemonize

brctl showmacs br0
port no mac addr is local? ageing timer
  1 00:15:99:e4:bb:ee no 28.58
  4 00:16:3e:0c:c7:70 no 20.44
  2 00:16:3e:1a:6b:53 no 4.14
  5 00:16:3e:5d:ee:52 no 0.77
  7 00:16:3e:6d:96:4c no 0.77
  6 00:16:3e:79:30:72 no 4.39
  2 00:ff:b8:4d:fb:98 no 0.22
  2 00:ff:d8:6e:6f:3f no 0.00
  7 02:85:2a:0e:23:61 yes 0.00
  6 12:32:ee:97:bf:99 yes 0.00
  2 2e:39:8b:0d:56:8d yes 0.00
  3 42:05:70:5b:70:2c yes 0.00
  5 46:01:af:8e:51:cc yes 0.00
  1 60:a4:4c:36:36:50 yes 0.00
  4 82:4f:d9:8c:fe:7a yes 0.00
  1 94:0c:6d:ec:7d:34 no 124.19

brctl showmacs br1
port no mac addr is local? ageing timer
  1 00:02:2a:f1:98:4b yes 0.00
  2 00:16:3e:3e:84:b6 no 0.00
  1 00:26:4d:5b:20:9c no 3.44
  1 00:30:88:1a:eb:e1 no 0.00
  2 56:74:0c:4c:73:6b yes 0.00

We found a Workaround to this situation but is a very bad solution.
When the conectivity problem exists, we conect via VNC to the WindowsXP VM and the conectivity is restored at once.
The VNC is connected via 192.168.2.50:5953 (via the HOST).

We tried to check the availability to connect to this port via nagios check_tcp and always is functioning (even when the conectivity is losed), so this is not functioning as a preventive measure.

Please, ask for any further information. Thanks.

[Hector Perez (heperez)]

Actually the Windows XP freeze. Is not only a comunication problem.

We can verify that when the problem occurs, the VM is freezed. Then, when we connect via VNC the machine continue working as never had been freezed, but the clock keep with the time of the freeze. We must adjust the clock.

[Serge Hallyn (serge-hallyn)]

Thanks for submitting this bug.

Would you be able to verify whether this is still the case when you use qemu from upstream git (git://git.qemu.org/qemu.git) ?

[Hector Perez (heperez)]

Hi Serge:

The server is in production now and i not familiar with git. I supposed that all VMs must be down before apply this command, is it?.
Would you please give me a guide how to do what you want?.
Thanks.

[Hector Perez (heperez)]

Ok, i have cloned the repository (with git clone git://git.qemu.org/qemu.git) but,
can i install to probe without destroy the previous install of qemu?, i am affraid to generate more problems with this version.
Thanks

[Serge Hallyn (serge-hallyn)]

Thanks Hector,

yes especially since you are running the VMs from the command line, you should be able to:

cd qemu #wherever the qemu git tree is
./configure --target-list=x86_64-softmmu
make

Then execute x86_64-softmmu/qemu-system-x86_64 -enable-kvm plus all your other options.

[Hector Perez (heperez)]

Ok,
We have created another WindowsXP VM, (a copy of the original one), and this copy have been executed with x86_64-softmmu/qemu-system-x86_64 -enable-kvm and all the parameters.

The VM started ok.
Now we have to wait a prudential period of time until we can say that the problem is solved, because the freeze problem occurs in an aleatory pattern.
Thanks, and we will inform the result.

[Hector Perez (heperez)]

Well, lamentably, with version 1.6.50 of qemu has become to repeat the problem. The VM guest is freezed again and it comed to life when whe had connected via VNC.

The problem isn't solved yet.