openstack team mailing list archive

[Vishvananda Ishaya <vishvananda@xxxxxxxxx>]

The api doesn't support any way to order these, so they are being returned in db order. In essex we retrieved each network individually, but now they are retrieved together. I believe the following will fix your issue:

diff --git a/nova/network/manager.py b/nova/network/manager.py
index d032fe1..4866803 100644
--- a/nova/network/manager.py
+++ b/nova/network/manager.py
@@ -1797,8 +1797,10 @@ class NetworkManager(manager.SchedulerDependentManager):
                                    project_only="allow_none")
 
     def _get_networks_by_uuids(self, context, network_uuids):
-        return self.db.network_get_all_by_uuids(context, network_uuids,
-                                                project_only="allow_none")
+        networks = self.db.network_get_all_by_uuids(context, network_uuids,
+                                                    project_only="allow_none")
+        networks.sort(key=lambda x: network_uuids.index(x['uuid']))
+        return networks
 
     @wrap_check_policy
     def get_vifs_by_instance(self, context, instance_id):
@@ -2114,8 +2116,10 @@ class VlanManager(RPCAllocateFixedIP, FloatingIP, NetworkManager):
         # NOTE(vish): Don't allow access to networks with project_id=None as
         #             these are networksa that haven't been allocated to a
         #             project yet.
-        return self.db.network_get_all_by_uuids(context, network_uuids,
-                                                project_only=True)
+        networks = self.db.network_get_all_by_uuids(context, network_uuids,
+                                                    project_only=True)
+        networks.sort(key=lambda x: network_uuids.index(x['uuid']))
+        return networks
 
     def _get_networks_for_instance(self, context, instance_id, project_id,
                                    requested_networks=None):

If you could verify that it helps and report a bug, I would be happy to propose it into essex and backport it.

Vish

On Nov 9, 2012, at 11:57 AM, Yi Sun <beyounn@xxxxxxxxx> wrote:

> I have two nic cards, I want eth0 to always be connected to network1
> and eth1 to always be connected to network2.
> In the essex, I can do this by arrange the order of --nic options in
> the "nova boot" command. But in the folsom,it does not seem to work
> anymore, the following commands connect eth0 to the same network
> nova --no-cache boot --image db64c7ed-3496-499e-a097-73362cb5f447
> --flavor 2 --nic net-id=ff899ad7-a0c2-4e46-a0a4-bf086eda9164 --nic
> net-id=12e964db-14e3-4e85-98c8-45987c89e1d1 user-host-2
> nova --no-cache boot --image db64c7ed-3496-499e-a097-73362cb5f447
> --flavor 2  --nic net-id=12e964db-14e3-4e85-98c8-45987c89e1d1 --nic
> net-id=ff899ad7-a0c2-4e46-a0a4-bf086eda9164 user-host-2
> 
> Could someone tell me how to solve this problem?
> Thanks
> Yi
> -- 
> Android-x86
> http://www.android-x86.org
> 
> _______________________________________________
> Mailing list: https://launchpad.net/~openstack
> Post to     : openstack@xxxxxxxxxxxxxxxxxxx
> Unsubscribe : https://launchpad.net/~openstack
> More help   : https://help.launchpad.net/ListHelp