Windows Server becomes unresponsive

savecancel

Hi Adrian, we do not have any tunneling setup in our environment./ Thanks.

can you put in a pastebin the output of the command "ps ax|grep kvm" from a kvm compute node? and what virtio drivers are you using?

Hi, I am using Red hat VirtIO driver(0.1.102) ver=03/10/2015,52.72.104.10200. Below is the result of the ps ax|grep kvm command. http://pastebin.com/jkM9stAb

We continue to face the same issue on the windows VM. Initially we found that the balloon drivers were not installed on affected VM and we installed it on all the VM. However this does not seem to have fixed the issue and few VM still becomes unresponsive after live migration.

You can try creating an image using the improved windows-openstack-imaging-tools with the new 0.1.126 stable virtio drivers, this might solve the issues you are having. Moreover, can you give a detailed view of your environment? As this is the first time you mentioned live migration issues. Thanks

hi Adrian, apologize to have been missed out mentioning live migration in actual post. Yes, we face this issue post live migration. We are using Openstack Juno with KVM and libvirt. I tested it with new drivers and still facing this issue. The VM having this issues are using QCOW2 images.

Hi, I might have found a possible reason for your issue. Can you try to run this powershell/cmd on the Windows instance before you migrate it:

powercfg.exe /S "8c5e7fda-e8bf-4a96-9a85-a6e23a8c635c"

Thank you,
Adrian Vladu

Hi Adrian, I tried changing the power option to high performance using above command and did a live migration, however the instance got stuck again after migration. I saw the below compute logs on the host that it was migrated.

DEBUG nova.virt.libvirt.driver [req-03d85ee3-04ba-4287-8ff4-d571138bbaaf None] [instance: ID] Could not determine fibre channel world wide node names get_volume_connector /usr/lib/python2.7/dist-packages/nova/virt/libvirt/driver.py:1282

DEBUG nova.virt.libvirt.driver [req-03d85ee3-04ba-4287-8ff4-d571138bbaaf None] [instance: ID] Could not determine fibre channel world wide port names get_volume_connector /usr/lib/python2.7/dist-packages/nova/virt/libvirt/driver.py:1289

Hi, We continue to face the same issue. Can QCOW2 images be the reason for this - we have boot from volume though in our environment.

By any chance, are you using same identical hardware for the kvm hypervisors nodes? This might be related to the exposed cpu capabilities for each node or some inconsistencies in the drivers.

Hi , Yes using the same identical hardware. Just came across th issue recently and found that the VM are getting paused, stopped , live migrated, started and resumed during the complete process of migration. below are the logs: http://pastebin.com/gkLYcMiv