Comment 1 for bug 794146

I've been able to resolve my issue by doing the following actions :
$ nova-manage project scrub $MY_PROJECT
$ nova-manage network delete 10.0.1.0/27
$ nova-manage network delete 10.0.1.0/26
$ nova-manage network create 10.0.1.0/24 1 256
$ service nova-network restart

=> select * from networks\G;
http://paste.openstack.org/show/1584/

My idea is that the instance were using a reserved ip by dnsmasq. After I scrub that range and affected a new one to the instance, it is now able to retrieve it's metadatas :

[...]
[ 0.707408] devtmpfs: mounted
[ 0.710454] Freeing unused kernel memory: 880k freed
[ 0.712021] Write protecting the kernel read-only data: 10240k
[ 0.713721] Freeing unused kernel memory: 84k freed
[ 0.722053] Freeing unused kernel memory: 1416k freed
lxcmount stop\/pre-start, process 45
cloud-init start-local running: Wed, 08 Jun 2011 08:27:53 +0000. up 1.93 seconds
no instance data found in start-local
init: cloud-init-local main process (167) terminated with status 1

init: udev-fallback-graphics main process (274) terminated with status 1

init: plymouth main process (47) killed by SEGV signal

init: plymouth-splash main process (277) terminated with status 2

init: cloud-init-nonet main process (228) killed by TERM signal

cloud-init start running: Wed, 08 Jun 2011 08:27:55 +0000. up 3.45 seconds
found data source: DataSourceEc2

[...]