[one-users] virtual network cannot get out

[Neil Schneider]

If I ping the gateway I see only
ARP, Request who-has  from the gateway/firewall for the network with no
answer
and
ICMP echo request from the vm with no response from the gateway/firewall.

So the firewall can't see the vm and the vm can't get responses from the
firewall.

When I try to ping my workstation from the VM I don't see packets entering
or leaving the em1 interface

I do however see ARP, Request who-has 172.16.168.184 tell 172.16.168.154
on vnet0 interface. 154 is the Virtual Machine, 184 is my workstation.

So it's sending ARP requests out the vnet interface but other machines are
ARPing to the em1 interface.

Any clues how I fix this?

Is this a problem with Open vswitch or an error created by Opennebula?

On Wed, May 7, 2014 2:26 am, Leszek Master wrote:
> Do you see any traffic from VM using tcpdump on em1?
>
>
> 2014-05-07 1:21 GMT+02:00 Neil Schneider <neil at ifxonline.com>:
>
>> I've been trying to work through this problem for two days and haven't
>> found the solution. I'm using opennebula to create virtual networks
>> using
>> openvswitch.
>>
>> [root at cloud1 ~]# ovs-vsctl show
>> c6def17d-2cc6-499e-a461-af4fe9aab78a
>>     Bridge management
>>         Port management
>>             Interface management
>>                 type: internal
>>         Port "vlan10"
>>             tag: 10
>>             Interface "vlan10"
>>                 type: internal
>>     Bridge public
>>         Port "em1"
>>             Interface "em1"
>>         Port "vnet0"
>>             Interface "vnet0"
>>         Port public
>>             Interface public
>>                 type: internal
>>     Bridge storage
>>         Port storage
>>             Interface storage
>>                 type: internal
>>         Port "vlan20"
>>             tag: 20
>>             Interface "vlan20"
>>                 type: internal
>>     ovs_version: "2.1.0"
>>
>> From the opennebula server I can see this.
>>
>> onevnet list
>>   ID USER         GROUP        NAME            CLUSTER      TYPE BRIDGE
>> LEASES
>>    0 oneadmin     oneadmin     management      ifx-produc      R
>> manageme
>>     0
>>    1 oneadmin     oneadmin     storage         ifx-produc      R storage
>>     0
>>    6 oneadmin     oneadmin     public          ifx-produc      R public
>>     1
>>
>> I've followed the instruction for configuring the hosting server so that
>> oneadmin has rights to access /var/lib/one on the hosting server as well
>> as sudo access to the scripts needed to create networks.
>>
>>
>> I have all the changes recommended to allow oneadmin to execute commands
>> through ssh to cloud1 the hosting server.
>>
>> oneadmin ALL=(ALL)      NOPASSWD: /usr/sbin/tgtadm, /sbin/lvcreate,
>> /sbin/lvremove, /bin/dd, /usr/bin/ovs-vsctl, /usr/bin/ovs-ofctl,
>> /usr/bin/ovs-dpctl, /sbin/iptables, /sbin/ebtables
>>
>> I can instantiate hosts from templates and everything works as expected.
>> When I bring up a virtual host, it gets an IP from the dhcp server
>> running
>> in the network. Not from the virtual network. Sorry, I can't cut and
>> paste
>> that part, since the only way I can access the virtual machine is
>> through
>> either VNC in sunstone or with virt-manager.
>>
>> I have another server running ovswitch that works fine. The main
>> difference is that I used virt-manager to create the hosts, instead of
>> opennebula. Those five virtual servers connect fine.
>>
>> [root at cloud2 ~]# ovs-vsctl show
>> aa56747f-d5a2-41b0-a998-48add3c62562
>>     Bridge public
>>         Port "vnet4"
>>             Interface "vnet4"
>>         Port "vnet0"
>>             Interface "vnet0"
>>         Port "vnet3"
>>             Interface "vnet3"
>>         Port public
>>             Interface public
>>                 type: internal
>>         Port "em1"
>>             Interface "em1"
>>         Port "vnet1"
>>             Interface "vnet1"
>>         Port "vnet2"
>>             Interface "vnet2"
>>     ovs_version: "2.1.0"
>>
>>
>> On cloud1 after the host gets it's IP address from the dhcp server
>> running
>> in our network, it can no longer connect to anything. I've checked
>> iptables rules, flushed them for testing, just to make sure. Everything
>> seems right, but the network isn't working.
>>
>> Sure would like to buy a clue. I've been searching the web for an answer
>> or an idea what to do to diagnose it. I suspect what's happening is that
>> opennebula/sunstone is not creating the interface properly. As I
>> understand the ip should be assigned to the bridge, not the virtual
>> interface.
>>
>> Sure could use some help. Even a pointer to a web site with the right
>> answer would be appreciated. I haven't been able to find it myself.
>>
>> Sorry for cross posting, but I couldn't decide which list to post to, so
>> I
>> did both.
>>
>> --
>> Neil Schneider                          pacneil_at_linuxgeek_dot_net
>>
>> This is your life. Do what you love, and do it often. If you don’t
>> like
>> something, change it. If you don’t like your job, quit. If you don’t
>> have
>> enough time, stop watching TV. If you are looking for the love of your
>> life, stop; they will be waiting for you when you start doing things you
>> love.”
>>
>> _______________________________________________
>> Users mailing list
>> Users at lists.opennebula.org
>> http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
>>
>