Dear experts of OpenvSwitch community,

Greetings!

I am a system admin who want to try out OpenvSwitch 1.10 on CentOS 6.x for 
using it for configuring a 2 node KVM redundant HA hypervisor cluster 
thereafter, and allowing VMs to migrate easily between the KVM
hosts.

I am stranded with confusion and lack of proper or concise document on open 
vswitch howto or config document.

Let me brief on my requirement:

I am trying to implement open source KVM (2 node HA cluster) on CentOS 6.x 
using OpenvSwitch 1.10 on a bond device.

I have installed CentOS 6.4, KVM and Open vSwitch 1.10 by building an rpm for 
my centos distribution, and installing it.

I have not bonded the physical interfaces on the CentOS based host machines 
yet, and intent to do so through openvswitch (ovs-vsctl command) by using the 
physical interfaces of the server (eth1 and eth2).

First I thought of creating a main bridge (cloudbr) and associate it with a 
bond device named as bond0 (which is non-existant until now) and specify that 
eth1 and eth2 are its associated interfaces.

Now, Do I need to define any Ports or Interfaces for this requirement?

Again, how do I configure IP address? and where to assign it?

I see that I can configure IP address using ifconfig on the main bridge 
(cloudbr) or the bond device (bond0) which I created by defining using the 
ovs-vsctl command. I also tried assigning IP address to any ports that I create 
(example: mgmt0) and associate any interface to it (example: mgmt0).

Then again, in none of these cases, it is not reachable or pingable...so I am 
missing something here?

Please find the commands that I tried to execute:

# ovs-vsctl add-br cloudbr -- add-bond cloudbr bond0 eth1 eth2 lacp=active 
other_config:lacp_time=fast -- set Bridge cloudbr fail-mode=secure -- set port 
cloudbr trunks=602,120,817 -- set bridge cloudbr stp_enable=true

(I have enabled LACP on the physical switches, and had tested connectivity to 
the ports without any problem)

# ifconfig cloudbr 192.168.120.100 netmask 255.255.255.0 broadcast 
192.168.120.255 up

Since the above command didnt work, I removed the IP address that was assigned, 
and tried the following as well:

# ovs-vsctl add-port cloudbr mgmt0 trunk=120,602,817 -- set interface mgmt0 
type=internal

# ifconfig mgmt0 192.168.120.100 netmask 255.255.255.0 broadcast 
192.168.120.255 up
OR
# ifconfig bond0 192.168.120.100 netmask 255.255.255.0 broadcast 
192.168.120.255 up

Still it is not working!

By assigning IP address by any of the above steps (ifconfig), I am not able to 
ping or reach the gateway or any external nodes from this server.

Kindly help with your guidance in resolving the issue and to understand the 
mistake in my approach!

I do not want to use any RedHat scripts or tools which one of the experts had 
recommended:

http://git.openvswitch.org/cgi-bin/gitweb.cgi?p=openvswitch;a=blob;f=rhel/README.RHEL

Instead I prefer to do it with generic commands as I intent to setup a open 
source KVM cluster (HA - 2 node) along with openvSwitch on CentOS 6.x. This 
will ensure that I learn all the aspects of configuring openvSwitch and how it 
really works.

Basically my intention is to eventually test all the options available with KVM 
like VM migration, HA, etc...and then integrate it with OpenStack.

thank you.

Kind Regards,
Ashraf Khalid

                                          
_______________________________________________
discuss mailing list
[email protected]
http://openvswitch.org/mailman/listinfo/discuss

Reply via email to