Network interface/dhcp problems ubuntu with kvm and network bridge

Question

I have a ubuntu system running 12.10

I am using KVM and have configured a bridge for the virtual machines, which is using nat forwarding. I am having problems with the connection dropping out periodically sometimes it stays up for a few days sometimes a few minutes.

I have disabled the bridge to resolve the issue but its still happening, I tried doing a manual dhcp release and renew which works sometimes, it gets me a new ip and im away again but i noticed that my ip is not updated when i run ifconfig or in kvm connection settings.

/etc/interfaces

auto lo
iface lo inet loopback

auto eth0
iface eth0 inet dhcp

#auto br0
#iface br0 inet dhcp
#    bridge_ports eth0
#    bridge_stp off
#    bridge_fd 0
#    bridge_maxwait 0

ifconfig

eth0      Link encap:Ethernet  HWaddr 38:60:77:aa:aa:aa  
      inet addr:10.0.0.137  Bcast:10.0.0.255  Mask:255.255.255.0
      UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
      RX packets:52068 errors:0 dropped:0 overruns:0 frame:0
      TX packets:10900 errors:0 dropped:0 overruns:0 carrier:0
      collisions:0 txqueuelen:1000 
      RX bytes:12322071 (12.3 MB)  TX bytes:2494188 (2.4 MB)

lo        Link encap:Local Loopback  
      inet addr:127.0.0.1  Mask:255.0.0.0
      UP LOOPBACK RUNNING  MTU:16436  Metric:1
      RX packets:2651 errors:0 dropped:0 overruns:0 frame:0
      TX packets:2651 errors:0 dropped:0 overruns:0 carrier:0
      collisions:0 txqueuelen:0 
      RX bytes:402096 (402.0 KB)  TX bytes:402096 (402.0 KB)

virbr0    Link encap:Ethernet  HWaddr f6:94:09:8f:7b:b5  
      inet addr:192.168.122.1  Bcast:192.168.122.255  Mask:255.255.255.0
      UP BROADCAST MULTICAST  MTU:1500  Metric:1
      RX packets:0 errors:0 dropped:0 overruns:0 frame:0
      TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
      collisions:0 txqueuelen:0 
      RX bytes:0 (0.0 B)  TX bytes:0 (0.0 B)

sudo dhcpclient -r -v

Internet Systems Consortium DHCP Client 4.2.4
Copyright 2004-2012 Internet Systems Consortium.
All rights reserved.
For info, please visit https://www.isc.org/software/dhcp/

Listening on LPF/virbr0/f6:94:09:8f:7b:b5
Sending on   LPF/virbr0/f6:94:09:8f:7b:b5
Listening on LPF/eth0/38:60:77:aa:aa:aa
Sending on   LPF/eth0/38:60:77:aa:aa:aa
Sending on   Socket/fallback
DHCPRELEASE on eth0 to 10.0.0.254 port 67

sudo dhcpclient -v

Internet Systems Consortium DHCP Client 4.2.4
Copyright 2004-2012 Internet Systems Consortium.
All rights reserved.
For info, please visit https://www.isc.org/software/dhcp/

Listening on LPF/virbr0/f6:94:09:8f:7b:b5
Sending on   LPF/virbr0/f6:94:09:8f:7b:b5
Sending on   Socket/fallback
DHCPDISCOVER on virbr0 to 255.255.255.255 port 67 interval 3
DHCPDISCOVER on virbr0 to 255.255.255.255 port 67 interval 5
DHCPDISCOVER on virbr0 to 255.255.255.255 port 67 interval 9
DHCPDISCOVER on virbr0 to 255.255.255.255 port 67 interval 15

At the moment the only way I know to recover from this state is to unplug the network cable and restart

UPDATE

When the connection drops out I have had some success running sudo dhclient -v without dropping the address first.

dhclient -v

Internet Systems Consortium DHCP Client 4.2.4
Copyright 2004-2012 Internet Systems Consortium.
All rights reserved.
For info, please visit https://www.isc.org/software/dhcp/

Listening on LPF/virbr0/96:a2:d6:f5:30:fc
Sending on   LPF/virbr0/96:a2:d6:f5:30:fc
Listening on LPF/eth0/38:60:77:aa:aa:aa
Sending on   LPF/eth0/38:60:77:aa:aa:aa
Sending on   Socket/fallback
DHCPDISCOVER on virbr0 to 255.255.255.255 port 67 interval 3
DHCPDISCOVER on eth0 to 255.255.255.255 port 67 interval 3
DHCPREQUEST of 10.0.0.129 on eth0 to 255.255.255.255 port 67
DHCPOFFER of 10.0.0.129 from 10.0.0.254
DHCPACK of 10.0.0.129 from 10.0.0.254
bound to 10.0.0.129 -- renewal in 5716 seconds.

here I can see I am being leased 10.0.0.129 but ifconfig still displays 10.0.0.137

eth0      Link encap:Ethernet  HWaddr 38:60:77:82:a8:61  
      inet addr:10.0.0.137  Bcast:10.0.0.255  Mask:255.255.255.0
      UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
      RX packets:312183 errors:0 dropped:0 overruns:0 frame:0
      TX packets:85529 errors:0 dropped:1 overruns:0 carrier:0
      collisions:0 txqueuelen:1000 
      RX bytes:80808972 (80.8 MB)  TX bytes:20479979 (20.4 MB)

lo        Link encap:Local Loopback  
      inet addr:127.0.0.1  Mask:255.0.0.0
      UP LOOPBACK RUNNING  MTU:16436  Metric:1
      RX packets:3910 errors:0 dropped:0 overruns:0 frame:0
      TX packets:3910 errors:0 dropped:0 overruns:0 carrier:0
      collisions:0 txqueuelen:0 
      RX bytes:461047 (461.0 KB)  TX bytes:461047 (461.0 KB)

virbr0    Link encap:Ethernet  HWaddr 96:a2:d6:f5:30:fc  
      inet addr:192.168.122.1  Bcast:192.168.122.255  Mask:255.255.255.0
      UP BROADCAST MULTICAST  MTU:1500  Metric:1
      RX packets:0 errors:0 dropped:0 overruns:0 frame:0
      TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
      collisions:0 txqueuelen:0 
      RX bytes:0 (0.0 B)  TX bytes:0 (0.0 B)

Solution

I'm genuinely surprised that no one has mentioned the obvious answer -

Drop DHCP and use static IP addressing. You can continue with NAT ETC, just don't assign the IP's from the DHCP pool, but in the same subnet and everything should work reliably and predictably.

All you need to do to enable static IP addressing is change

auto eth0
iface eth0 inet dhcp

to

auto eth0
address X.X.X.X
iface eth0 inet static
netmask 255.255.255.0  
gateway X.X.X.Y

Making sure the gateway is the address assigned by KVM (if you do a /sbin/route -n it will show you the default gateway - thats the IP address to use).

Comments

[oo.]

What type of network card do you have?

[Alex Edwards]

lspci | egrep -i --color 'network|ethernet' 05:00.0 Network controller: Ralink corp. RT5390 Wireless 802.11n 1T/1R PCIe 06:00.0 Ethernet controller: Realtek Semiconductor Co., Ltd. RTL8111/8168B PCI Express Gigabit Ethernet controller (rev 06)

[Alex Edwards]

maybe I should disable that wireless card as i'm not using that

[rubo77]

can you post the output of tail -f /var/log/syslog? and a tcpdump of networkmanager while the connection gets lost? (best on pastebin.com )

[Alex Edwards]

@rubo77 will do, it could be a while its been well behaved today!

[Alex Edwards]

well I have disabled the bridge which is br0, when this is enabled eth0 has no ip it appears on br0, but I still get the same connection dropout issues. virbr0 is something kvm created, I think this is the nat for the virtual hosts which get externally addressable ips when they are running but this config is without any vm's running.

[killermist]

Ya. It was a shot in the dark, and far too long to be a comment to the question. I figured it might spark some ideas that might lead you to the correct result. I hope you figure it out, or that someone else knows the answer.

[Alex Edwards]

ipv6 should already be disabled I have net.ipv6.conf.all.disable_ipv6=1, net.ipv6.conf.default.disable_ipv6=1, net.ipv6.conf.lo.disable_ipv6=1, in my sysctl.conf file

[Thufir]

good static ip explanation for bridging