← Back to team overview

openstack team mailing list archive

Re: OpenStack Multi Node Problems with OVS

 

It is Private Networks (VLAN).  Networking works between all the devices.  I am pretty sure this is a bug.  I am also seeing the following message on my compute server (see below).
The error message is indicating that it is not getting the proper values when running the command.  I confirmed as user root, quantum, nova that I can easily run the command get
the output.

Atif...

==========================================

 /etc/quantum/plugins/openvswitch/ovs_quantum_plugin.ini

[OVS]
tenant_network_type=vlan
network_vlan_ranges = physnet1:500:1000
bridge_mappings = physnet1:br-eth1
===========================================

openvswitch-agent.log


2013-04-17 07:29:37    DEBUG [quantum.agent.linux.utils] Running command: sudo /usr/bin/quantum-rootwrap /etc/quantum/rootwrap.conf ovs-vsctl --timeout=2 list-ports br-int

2013-04-17 07:29:38    DEBUG [quantum.agent.linux.utils] 
Command: ['sudo', '/usr/bin/quantum-rootwrap', '/etc/quantum/rootwrap.conf', 'ovs-vsctl', '--timeout=2', 'list-ports', 'br-int']
Exit code: 0



________________________________
 From: Robert van Leeuwen <Robert.vanLeeuwen@xxxxxxxxxxxxx>
To: "openstack@xxxxxxxxxxxxxxxxxxx" <openstack@xxxxxxxxxxxxxxxxxxx> 
Sent: Wednesday, April 17, 2013 3:45 AM
Subject: Re: [Openstack] OpenStack Multi Node Problems with OVS
 


 
> Hi Folks, 
>
> I am working on bringing up a multi node OpenStack environment with OVS.  
> I have a Controller, Compute and a Gateway/Network node.  
> This is running Folsom.  Most of the services are up, except that I cannot ping the floating ip of the VM.  

What kind of setup are you creating? Bridge_mapped networks? Private networks (with GRE tunnels)?

If you are setting this up with GRE I'm  missing the GRE tunnels in the "ovs-vsctl show" overview.
You should have GRE tunnels to all Compute nodes in the network.

To start troubleshooting I would first check if networking works between compute nodes:
Start two VM's, on different Compute nodes, in the same subnet and see if they can reach each other.
(If they do not get a DHCP IP just put in a static IP for now)

There is bug with rebooting the machines running dhcp / l3-agent which might be good to be aware of:
https://bugs.launchpad.net/quantum/+bug/1091605

Cheers,
Robert van Leeuwen

_______________________________________________
Mailing list: https://launchpad.net/~openstack
Post to     : openstack@xxxxxxxxxxxxxxxxxxx
Unsubscribe : https://launchpad.net/~openstack
More help   : https://help.launchpad.net/ListHelp

References