After investigation virtual machines off the network trigger physical machine migration strategy, then moved back from node migrated unavailable
Experiment:
- Node1 physical machines running virtual machines vm_test
- Unplug the network cable node1 physical machine, using virsh list --all view a running virtual machine vm_test
- Wait for the virtual machine to a physical machine migration vm_test node2
- Virtual machine migration to node2 success vm_test
- were in the physical machine node1, use node2 virsh list --all view a running virtual machine,At this time, vm_test exist in the physical machine node1 and node2, and run simultaneously.Then again vm_test want to migrate from node2 back to node1, certainly fail.(Performing dynamic migration, is the name of the client migration is unique on the host does not have a purpose and source host are migrated clients with the same name exist in the client)
- Manual destroy physical machine node1 virtual machine vm_test, again migrating from a physical machine to node1 node2 vm_test, successful migration.
Issue:
I have read the code, after pulling off the network cable is redeployment, rather than migration,Why opennebula not designed to migrate off the net rather than the redeployment?