openstack team mailing list archive

[Lucio Cossio <lucio_cossio@xxxxxxxxxxxx>]

Hello Guys, I'm testing a dual node installation of OpenStack Nova with
Glance, and i really hope someone can help me with my current problem.

My setup is like that:
- First Node : All nova components plus Glance
- Second Node : nova-compute

I'm using diablo version  that was installed from the ubuntu repository,
which i suppose is the nova 2011.3+git20111117-0ubuntu1 oneiric version.

For what a read, the second node only needs nova-compute ,but sometimes i
see others saying about using nova-network together. I'm using machines
that have just one network interface, so they are in the same network with
other non-OpenStack computers (this network uses dhcp).
At the network configuration i'm using flatDHCP. The nova.conf file can be
see here (the ip is not exactly what im using, is just a template):
http://paste.openstack.org/show/3978/

So, i'm able to install and run virtual machines in both nodes. My problem
is, for any instance that is running on the second node, i can't connect
with the outside world (i can ssh and ping it from the first node, but not
from an outside computer). Into the first node all goes fine.

If i restart nova-network (not exactly one time, sometimes more) at some
point nova-network goes down, and then i'm able to ping the instances. Is
like the configuration is good but the service break. The error that
appears on logs is Stderr: 'RTNETLINK answers: File exists\n' -
http://pastebin.com/qYhhYS6h

Currently i'm trying the same configuration in another 2 computers, using
the diablo-final version of the rcb repository, as the openstack forum
suggest. The errors is different. From what i remember (i'm not the only
working with that) i was not even able to ping instances from outside after
restarting the nova-network.
Then we tried to use nova-network with the second node too. After some
time, new instances running on the second node start gracefully, and i can
ping them from outside, however i can't run instances on the first node
anymore. There's no log errors at nova or qemu folder, i can vnc to the
instance but it appears stopped (cant do nothing).

I appreciate any help