Description of problem: I have 3.5 SHE EL6 host with engine 3.6.8 and I added on EL7 3.6 host into Default cluster which is in InClusterUpgrade scheduling policy. I tried to migration a VM (HE VM, while being in global maintenance) from SHE EL6 host to "normal" EL7 host (not part of SHE as deployment was not successful). Migration attempt ends in error as: ~~~ Operation Canceled Error while executing action: HostedEngine: Cannot migrate VM. Selected display type is not supported by the operating system. ~~~ This HE VM used to be 3.4 -> 3.5 -> 3.6. Thus maybe there's something "forgotten" from the last in VM's configuration. Version-Release number of selected component (if applicable): rhevm-backend-3.6.8.1-0.1.el6.noarch How reproducible: 100% Steps to Reproduce: 1. have 3.4 -> 3.5 -> 3.6 engine HE VM (running on a 3.5 EL6 host) 2. add 3.6 EL7 host into Default cluster (not SHE host) 3. have InClusterUpgrade scheduling policy defined 4. migration HE VM (global maintenance) Actual results: Cannot migrate VM. Selected display type is not supported by the operating system. Expected results: Should probably work or at least the forbidden action should not be related to display type. Additional info:
[RHEVM shell (connected)]# show vm HostedEngine id : 176b8a31-a703-443a-ad11-9aef06f0349e name : HostedEngine bios-boot_menu-enabled : False cluster-id : 00000001-0001-0001-0001-000000000229 cpu-architecture : X86_64 cpu-topology-cores : 1 cpu-topology-sockets : 2 cpu-topology-threads : 1 cpu_profile-id : 57f6173a-e515-438a-b2ad-82ba79343e18 cpu_shares : 0 creation_time : 2016-08-04 15:13:05.734000+02:00 delete_protected : False display-address : 10.34.60.151 display-allow_override : False display-copy_paste_enabled : True display-disconnect_action : LOCK_SCREEN display-file_transfer_enabled: True display-monitors : 1 display-port : 5900 display-single_qxl_pci : False display-smartcard_enabled : False display-type : vnc high_availability-enabled : False high_availability-priority : 0 host-id : 548245b1-b839-496e-97ae-8b06e1ee8bf9 io-threads : 0 large_icon-id : 6c46191a-4eda-401a-823f-b40e94727d4c memory : 4294967296 memory_policy-guaranteed : 4294967296 migration-auto_converge : inherit migration-compressed : inherit migration_downtime : -1 next_run_configuration_exists: False numa_tune_mode : interleave origin : hosted_engine os-boot-dev : hd os-type : other placement_policy-affinity : migratable run_once : False small_icon-id : 2a490ccd-b976-40bb-93f6-812a8c006f51 sso-methods-method-id : GUEST_AGENT start_paused : False stateless : False status-state : up stop_time : 2016-08-05 17:14:11.998000+02:00 template-id : 00000000-0000-0000-0000-000000000000 time_zone-name : Etc/GMT timezone : Etc/GMT type : desktop usb-enabled : False
I compared this HE VM with another HE VM in different setup (which also is VNC) and it had different value in it's engine DB. Thus following workaround made the HE VM migrate. UPDATE vm_static set default_display_type = 1 where vm_name = 'HostedEngine'; and restart of engine.
This seams a side effects of rhbz#1339305.
For now this is not a supported case. You should have all hosts migrated to 3.5/el7 which allows upgrade to 3.6/el7 without further mixing versions and operating systems. Please see the docs.
This could be related to that: https://gerrit.ovirt.org/#/c/58314/3/packaging/dbscripts/upgrade/03_06_2130_add_graphics_and_video_devices_to_he.sql As far as I remember, to make evertying work on 3.6 cluster level I had to set it to default_display_type = 0. Maybe this introduces a problem on 3.5 cluster level.
Hello, Today, I'm trying to export from 3.6 then import into 4.3, and I'm stuck with the same issue. Thanks to the hint of the workaround #c3, I modified the raw nfs_export_domain_server:/nfs_shared_dir/xxx/master/vms/yyy/yyy.ovf and found <DefaultDisplayType>0</DefaultDisplayType> and changed it to 1. Now, the import action does not yell any display error.