Bug #3769

on KVM node live migrate should not start if there are disks with attribute CACHE!=none

Added by Anton Todorov about 6 years ago. Updated about 6 years ago.

Status:ClosedStart date:04/21/2015
Priority:NormalDue date:
Assignee:-% Done:

0%

Category:Core & System
Target version:-
Resolution:worksforme Pull request:
Affected Versions:OpenNebula 4.10, OpenNebula 4.12

Description

Live migrate should be disabled at least for KVM node if there are disks attached to VM with CACHE!=none
Here is the log:

Tue Apr 21 13:13:35 2015 [Z0][DiM][D]: Live-migrating VM 5
Tue Apr 21 13:13:35 2015 [Z0][ReM][D]: Req:3616 UID:0 VirtualMachineMigrate result SUCCESS, 5
Tue Apr 21 13:13:35 2015 [Z0][ReM][D]: Req:9920 UID:0 VirtualMachineInfo invoked , 5
Tue Apr 21 13:13:35 2015 [Z0][ReM][D]: Req:9920 UID:0 VirtualMachineInfo result SUCCESS, "<VM><ID>5</ID><UID>0..." 
Tue Apr 21 13:13:37 2015 [Z0][ImM][I]: --Mark--
Tue Apr 21 13:13:38 2015 [Z0][VMM][D]: Message received: LOG I 5 Successfully execute transfer manager driver operation: tm_premigrate.
Tue Apr 21 13:13:38 2015 [Z0][VMM][D]: Message received: LOG I 5 ExitCode: 0
Tue Apr 21 13:13:38 2015 [Z0][VMM][D]: Message received: LOG I 5 Successfully execute network driver operation: pre.
Tue Apr 21 13:13:39 2015 [Z0][VMM][D]: Message received: LOG I 5 Command execution fail: /var/tmp/one/vmm/kvm/migrate 'one-5' 's05' 's06' 5 s06
Tue Apr 21 13:13:39 2015 [Z0][VMM][D]: Message received: LOG E 5 migrate: Command "virsh --connect qemu:///system migrate --live  one-5 qemu+ssh://s05/system" failed: error: Unsafe migration: Migration may lead to data corruption if disks use cache != none
Tue Apr 21 13:13:39 2015 [Z0][VMM][D]: Message received: LOG E 5 Could not migrate one-5 to s05
Tue Apr 21 13:13:39 2015 [Z0][VMM][D]: Message received: LOG I 5 ExitCode: 1
Tue Apr 21 13:13:39 2015 [Z0][VMM][D]: Message received: LOG I 5 Failed to execute virtualization driver operation: migrate.

ps. I am marking this as 'Core & System' because I do not know how to define that both shell and sunstone should be fixed

History

#1 Updated by Carlo Daffara about 6 years ago

For kvm, if a clustered file system is used, all cache modes support live migration. For example, see here:
https://www.suse.com/documentation/sles11/book_kvm/data/sect1_4_chapter_book_kvm.html

#2 Updated by Anton Todorov about 6 years ago

Thank you for the answer

Missed this part of the documentation too:
[[http://docs.opennebula.org/4.12/administration/virtualization/kvmg.html#considerations-limitations]]

I do realize that there are too many libvirt versions on the different OSes and it is hard to detect when to rise alert or no.

I think the issue should be closed.

#3 Updated by Ruben S. Montero about 6 years ago

  • Status changed from Pending to Closed
  • Resolution set to worksforme

Also available in: Atom PDF