With bridged networking you can share actual network device with KVM machines. This is required for servers with multiple network cards and gives you good performance. You can choose to put multiple segments into one bridged network or to divide it into different networks interconnected by routers.
Our Sample Setup
The following describes the networking used by our setup:
+------+ +-----+
LAN --> | eth0 | | br0 | -> KVM VMs connected to LAN for SAN/NFS
+------+ +-----+
10.10.x.y/24
+------+ +-----+
Internet --> | eth1 | | br1 | -> KVM VMs connected to the Inernet
+------+ +-----+
123.1.2.0/28
Where,
- All other clients can reached to all VMs via br1 which is connected to public interface. br1 is our default gateway.
- br0 is connected to private LAN to access other servers, services and storage devices such as SAN/NAS or NFS servers. br0 route is configured via route-br0 static networking configuration file.
Turn Off NetworkManager
The NetworkManager (GUI) tool can create problems with bridged based networking so disable it as follows, enter:
# chkconfig NetworkManager off
# chkconfig network on
# service NetworkManager stop
Edit /etc/sysconfig/network, enter:
# vi /etc/sysconfig/network
Update file as follows:
NETWORKING=yes HOSTNAME=kvm42.nixcraft.net GATEWAY=br1
Save and close the file.
br0: Configure Bridging for eth0
Edit /etc/sysconfig/network-scripts/ifcfg-eth0, enter:
# vi /etc/sysconfig/network-scripts/ifcfg-eth0
Update it as follows:
DEVICE=eth0 ONBOOT=yes ETHTOOL_OPTS="autoneg off speed 100 duplex full" BRIDGE=br0 HWADDR=b8:ac:6f:65:31:e5
Save and close the file. Create /etc/sysconfig/network-scripts/ifcfg-br0, enter:
# vi /etc/sysconfig/network-scripts/ifcfg-br0
Update it as follows (note options are case sensitive i.e. Bridge and bridge are two different options)
DEVICE=br0 TYPE=Bridge BOOTPROTO=static ONBOOT=yes IPADDR=10.10.21.70 NETMASK=255.255.255.192 DELAY=0
Save and close the file. Finally, create static routing file /etc/sysconfig/network-scripts/route-br0, enter :
# vi /etc/sysconfig/network-scripts/route-br0
Edit it as follows:
10.0.0.0/8 via 10.10.21.122 dev br0
Save and close the file. Delete old /etc/sysconfig/network-scripts/route-eth0 (if exists):
# rm /etc/sysconfig/network-scripts/route-eth0
br1: Configure Bridging for eth1
Edit /etc/sysconfig/network-scripts/ifcfg-eth1, enter:
# vi /etc/sysconfig/network-scripts/ifcfg-eth1
Update it as follows:
DEVICE=eth1 ONBOOT=yes ETHTOOL_OPTS="autoneg off speed 100 duplex full" BRIDGE=br1 HWADDR=00:30:49:8c:48:ad
Please note that test server is set to 100Mbps full duplex. Save and close the file. Edit /etc/sysconfig/network-scripts/ifcfg-br1, enter:
# vi /etc/sysconfig/network-scripts/ifcfg-br1
Define public IP address including gateway as follows:
DEVICE=br1 BOOTPROTO=static ONBOOT=yes IPADDR=123.1.2.3 NETMASK=255.255.255.248 GATEWAY=123.1.2.200 TYPE=Bridge DELAY=0
Save and close the file.
Restart Network Service
Type the following command:
# service network restart
Make sure everything is working fine:
# brctl show
Sample outputs:
bridge name bridge id STP enabled interfaces br0 8000.0030488e31ac no eth0 br1 8000.0030488e31ad no eth1 virbr0 8000.000000000000 yes
Verify IPs and routing tables:
# ip addr show br0
# ip addr show br1
# ip route
# ping cyberciti.biz
# host google.com
- CentOS / Redhat: Install KVM Virtualization Software
- CentOS / Redhat: KVM Bridged Network Configuration
- KVM virt-manager: Install CentOS As Guest Operating System
- KVM virt-install: Install FreeBSD / CentOS As Guest Operating System
- KVM: Install CentOS / RHEL Using Kickstart File (Automated Installation)
- Troubleshooting KVM Virtualization Problem With Log Files
- KVM Virsh: Redirect FreeBSD Console To A Serial Port
- KVM: Starting / Stopping Guest Operating Systems With virsh Command
- Linux KVM: Disable virbr0 NAT Interface
- FreeBSD / OpeBSD Running in KVM Does Not Accept FTP Traffic
- KVM: Start a Virtual Machine / Guest At Boot Time
- KVM virt-install: Install OpenBSD As Guest Operating System
- Linux KVM: OpenBSD Guest Hangs At Starting tty Flags
- KVM Virtualization: Start VNC Remote Access For Guest Operating Systems
🐧 Get the latest tutorials on Linux, Open Source & DevOps via:
- RSS feed or Weekly email newsletter
- Share on Twitter • Facebook • 25 comments... add one ↓
| Category | List of Unix and Linux commands |
|---|---|
| File Management | cat |
| Firewall | Alpine Awall • CentOS 8 • OpenSUSE • RHEL 8 • Ubuntu 16.04 • Ubuntu 18.04 • Ubuntu 20.04 |
| Network Utilities | dig • host • ip • nmap |
| OpenVPN | CentOS 7 • CentOS 8 • Debian 10 • Debian 8/9 • Ubuntu 18.04 • Ubuntu 20.04 |
| Package Manager | apk • apt |
| Processes Management | bg • chroot • cron • disown • fg • jobs • killall • kill • pidof • pstree • pwdx • time |
| Searching | grep • whereis • which |
| User Information | groups • id • lastcomm • last • lid/libuser-lid • logname • members • users • whoami • who • w |
| WireGuard VPN | Alpine • CentOS 8 • Debian 10 • Firewall • Ubuntu 20.04 |
A small typo needs to be corrected Under “br1: Configure Bridging for eth1” heading, “vi /etc/sysconfig/network-scripts/ifcfg-eth0” should be changed with “vi /etc/sysconfig/network-scripts/ifcfg-eth1”
Thanks for the heads up! The post has been updated.
regarding the Bridging…
watch the custom options ( ETHTOOL_OPTS ) you pass in eth0, eth1 etc !!
you may find network issues in guest machines. I have three guest ( Ubuntu,Centos, Openbsd ) in my Centos host and the 3 guest had problems.Probably has to do with the model of the network card..so leave the default!!!
something like this :
DEVICE=eth0
ONBOOT=yes
BRIDGE=br0
HWADDR=XX:XX:XX:XX:XX:XX
How would a single eth0 nic grabbing its information from a DHCP server be handled?
I think your GATEWAY parameter in /etc/sysconfig/network is incorrect. That parameter should be the IP address. If you want to direct to a device, then you need to use GATEWAYDEV parameter.
http://www.centos.org/docs/5/html/Deployment_Guide-en-US/ch-sysconfig.html
Thanks for the good work on your site. I always learn something new.
Dear Vivek,
Can U please elaborate on this entry.
10.0.0.0/8 via 10.10.21.122 dev br0
I also do not know why you would need a static route. This route should be included by default.
The RHEL docs do not mention this.
How if eth1 have 2 class ip ?
I got up to the point where you have to edit
/etc/sysconfig/network-scripts/ifcfg-eth0
I dont have that file….
Its a clean install too….
Ok how do I remove the bridge?
Delete ifcfg-brX and restart the network, make sure you restore your route-eth0 file if you’re removing the private network bridge
Actually i am using wireless to connect to my modem (ifcfg-wlan0) from my host machine. if i create bridge as br0, in my host machine, will there be any change in ifcfg-br0 for wireless setup.
Also, whether i need to include type=bridge parameter in ifcfg-wlan0 file of my host machine?
Wheter i need to use eth0 or wlan0 in my kvm for connecting this bridge.
What if I have only one nic which is connected (wired) to a wireless router and internet gateway and I want my guest VMs accessible from the LAN and can access the internet as well?
Appreciating your support.
Don’t try this on a remote production machine with a public IP……you’ll lose connectivity. I haven’t found a guide for bridging anything yet that won’t cause a production machine to go down :(
My RHEL 5.8 box has one nic with two static IP’s (one VIP) – I have a DHCP server in my environment – do I need bridging?
How do you map your VM to use br0 ? When adding your VM using virt-manager, which network should my VM card be added to ?
If I add to virbr0, i get mapped to the 192.168.x.x/24 private network.
I can’t seem to create a new network with the same network as br0 cause it doesn’t allow it.
Any help appreciated.
This is so idiotic. Doing things round about way in redhat KVM virualization. Just use Oracle’s Virtual box. So much easier and not at all confusing.
@ CheapIndianITGuy: Why would anyone do this in Oracle virtualbox if their goal was to be an RHCSA or RHCE???? It defeats the purpose of developing a level of proficiency with RedHat KVM (a “known” RH Linux test objective) as well as gaining an overall familiarity of the OS’s functionality. Heck, even a linux enthusiast can gain from learning this method.
Personally, the guy who delves a bit deeper into the woodworks has more than a “job;” he has a “career…”
Thanks for the good work friends..
I am also having bit issue in my KVM host machine.
1- I have one eth0 interface in my HOST Machine.
2- KVM Installed on RHEL 6.0 and XP is running as Guest
I try making bridge from KVM GUI interface with eth0 for my KVM guests to use internet . BUt first it hangs up..i forcefully close the GUI of making bridge with eth0 and then I try remove “rm /etc/sysconfig/network-scripts/ifcfg-br0”
File is removed but i didn’t get my eth0 back to normal When i use ifup eth0,it gives me:
:: ERROR ::
SIOCGMIIPHY on ‘eth0’ failed: Resource temporarily unavailable SIOCGMIIPHY
Please i am not very experienced but i need help in this regard, ??
Really thankful friends if you just help me in this
Nice write-up. You apparently know what you are doing, so while I plan to study your post carefully and see if I can solve my own question(s) later tonight, I thought I would post my target topology and perhaps someone can let me know what is and isn’t possible, and perhaps point me in the right direction. Thanks in advance for any help!
My host platform has 4 GigE ports and is running CentOS 6.3 with KVM virtualization. I will have three different VMs running — two will probably be Windows 7 and one will be RHEL. What I want to accomplish is to assign a separate, independent GigE port to each of the VMs. In a picture:
eth0 Host System
eth1 VM1
eth2 VM2
eth3 VM3
The Host system will not use eth1, eth2, or eth3 *at all* and should not respond to any traffic on them. Similarly, each VM is oblivious to the traffic on the three ports that are not assigned to it. And if a VM needs to talk to different VM, the traffic should leave the platform via one port, be routed by the external network(s) and fed back in via a different port (assuming a route exists). [and by ‘external network’, I mean outside the VM Host, not outside the building]
I know how to use ONE port to give each VM its own connection to the external network with unique MAC and IP Addresses, so I suppose I could implement a similar scheme where each VM gets its own connection via a different physical port, but how do I tell the host itself to keep its hands off that traffic.
I am convinced there is a simple solution to this but I think I keep getting caught up in unnecessary complications (and confusions).
Thanks in advance!
How i can add to NICs to KVM machine 1 for private and the 2 for public ?
is it mandatory to assign IP on bridge device ? Can we directly assign public IP to guest vm ?
I’m trying to set up a KVM host with multiple NICs in a slightly different manner. I have two integrated broadcom nics on the motherboard and a PCIe board with 2x 10000Gbe fiber ports.
The ixgbe port I am using works fine when I assign an IP, but I want it 802.1Q tagging to work between the local fiber port and the switch.
The intent is to let the host communicate on eth0 for its own management and connectivity, and all the guest VMs to have a local static IP and a connection to a virtual device (bridge, I believe) talking on the appropriate VLAN.
I think I need to create multiple virtual-bridge interfaces, one for each VLAN, but I’ve not found this clearly documented anywhere yet.
Any ideas?
Great howto but if my local lan is 192.168.0.0/24 how should the static route look ?
I have Centos 6 box installed with a few KVM machines. My question is how to add bulk IPs to a specific VM without doing it manually via SolusVM control panel.
I followed your guide, but I think this is more specific for multiple NIC cards. Would appreciate if you can explain how one can add bulk IPs to a KVM VM on a Centos 6x NODE.
I already have the first 3 steps set up accordingly.
/etc/sysconfig/network
/etc/sysconfig/network-scripts/ifcfg-eth0
/etc/sysconfig/network-scripts/ifcfg-br0