Hi , I deployed both of Icehouse and Havana OpenStack version which have 1 Controller Node, 1 Network Node and 15 Compute Nodes connected together via VPN network so i want to resize instance at same node because migration to other node take more time. I've already edit allow_resize_to_same_host = True and scheduler_default_filters = AllHostsFilter in both of Controller Node and Compute Nodes. I faced the problem of Icehouse and Havana

Problem in OpenStack Havana when i try to resize instance, sometime OpenStack try to migrate instance to other nodes and sometime instance can be resized successfully at same node. I'm sure my current node that instance is located have enough resources. why does OpenStack sometime try to migrate after i allow_resize_to_same_host ? moreover, nova-compute.log just show error about ssh to other nodes and my point just want to resize at same node. It should not migrate to other nodes anymore.

Problem in OpenStack Icehouse (Compute01 is a first Compute Node that be installed before other Compute Nodes) when i try to resize instance which is not launched to Compute01 in case instance is launched between Compute02-15, OpenStack will migrate instance to Compute01 every time. On the other hand, if instance is launched to Compute01, it can be resized successfully. I think Icehouse is difference from Havana, because Compute01's instance can be resized successfully every time.

I want to force resize instance to the same node only. I would like to know which OpenStack python files that involved with this problem? For me, i deeply found some file in /usr/lib/python2.7/dist-packages/nova/compute/ such as api.py and manage.py which has resize_instance method. Finally, I just want to know How to modified code of OpenStack resize instance both of havana and icehouse ?

Best regards, totten25