[one-users] Creating small networks with XMLRPC

[Daniel Morán Jiménez]

Hello,

Testing the network creation with XMLRPC I have met a problem with small
subnets, and I think it may be because of my lack of knowledge about how the
SIZE parameter of the virtual network template is interpreted. I have tried
to deploy the following networks 10.95.240.16/28 and 10.95.240.8/29.  Once
created, I have deployed a virtual machine with two network adapters, one in
each of the network, and I found that OpenNebula gives the same lease
10.95.240.1 (and the same MAC address) for each adapter. That's obviously
wrong, not only because of it being duplicated, but also because the lease
doesn't belong to any of the networks. I think the problem is my assumption
of the SIZE parameter being used to calculate the type of the network. How
is the network mask calculated for each network? After a few tests, the
network lease mechanism seems to work fine for networks equal or bigger than
class C. Are the sizes of the networks calculated as classful networks?

The strange thing is that I only get this problem when the network is
created with the XMLRPC API; if I create the networks with templates and
command-line tools before deploying the machines the error is not thrown. Is
it any difference between creating networks with XMLRPC and creating them
with the command line tool? Does the network creation behavior change if the
network is created with different users (I use different users for
command-line and XMLRPC calls)? Using "onevnet show" in the networks created
both ways, and comparing the results, these seem to be identical, but the
leasing mechanism doesn't work with the XMLRPC networks for this particular
cases.

Thanks in advance,

Regards
Daniel
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opennebula.org/pipermail/users-opennebula.org/attachments/20100119/0f17f14e/attachment-0002.htm>