Bug 2152909

Created attachment 1932370 [details]
virshdump

Description of problem:

We want to migrate VM from Node-to-Node, but it is failing with below error.

Linux:
VirtualMachineInstance migration uid 29acdc88-b8d7-4ab2-add2-1131a6d8868a failed. reason:Live migration failed error encountered during MigrateToURI3 libvirt api call: virError(Code=1, Domain=10, Message='internal error: unable to execute QEMU command 'cont': Failed to get "write" lock')

Windows:
server error. command SyncVMI failed: "LibvirtError(Code=1, Domain=10, Message='internal error: unable to execute QEMU command 'cont': Failed to get \"consistent read\" lock')"


> I am using Ceph(not RedHat Ceph) as my storage class with RWX mode.

> now, if used node selector in my YAML, VM runs on the particular node without any error.

> On initiating node to node migration from UI, it creates 1 more virt launcher pod in another node, but it goes on into completion rather than continue to stay in running. Then, migration fails with above error.

> In the second virt-launcher pod, on doing virsh list, we don't find any VMs running.

> Exisitng virt-launcher pod, we observe 

2022-12-13 13:14:22.884+0000: initiating migration
2022-12-13T13:14:25.532764Z qemu-kvm: warning: Failed to unlock byte 201
2022-12-13T13:14:25.532834Z qemu-kvm: warning: Failed to unlock byte 201


Version-Release number of selected component (if applicable):

4.10.6

How reproducible:

75%


Actual results:

Node-to-Node Live Migration should be completed without any error.


Additional info:

Attach virsh dump xml, libvirtd, virt-launcher logs

>
virsh # list
 Id   Name                                    State
------------------------------------------------------
 1    migration_migration-test-10-1-100-181   paused

virsh # resume migration_migration-test-10-1-100-181
error: Failed to resume domain 'migration_migration-test-10-1-100-181'
error: Timed out during operation: cannot acquire state change lock (held by monitor=remoteDispatchDomainGetBlockInfo)