-
Bug
-
Resolution: Unresolved
-
Undefined
-
CentOS Stream 9, rhel-9.5
-
None
-
None
-
sst_virtualization
-
ssg_virtualization
-
5
-
False
-
-
None
-
Red Hat Enterprise Linux
-
None
-
None
-
None
-
-
x86_64
-
None
What were you trying to do that didn't work?
Trying to live migrate a paused domain with block migration back and forth between two hypervisors.
I'm using:
- OSP 18.0 (pre-GA) with openstack-nova 2023.1
- libvirt-10.4.0-1.el9.x86_64
- qemu-kvm-9.0.0-6.el9.x86_64
- CentOS Stream release 9
The hypervisors are L1 VMs top of a single L0 machine that uses nested virt with:
- qemu 9.0.1
- libvirt 10.5.0
The first live migration succeeds, but the live migration moving the domain back to the starting hypervisor fails with the qemu process crashing:
source hypervisor:
2024-07-16T13:28:29.971381Z qemu-kvm: warning: Deprecated CPU topology (considered invalid): Unsupported clusters parameter mustn't be specified as 1
warning: old compression is deprecated; use multifd compression methods instead
warning: old compression is deprecated; use multifd compression methods instead
warning: old compression is deprecated; use multifd compression methods instead
warning: block migration is deprecated; use blockdev-mirror with NBD instead
2024-07-16 13:28:32.608+0000: Domain id=1 is tainted: custom-monitor
2024-07-16 13:29:55.375+0000: initiating migration
qemu-kvm: ../block.c:6979: int bdrv_inactivate_recurse(BlockDriverState *): Assertion `!(bs->open_flags & BDRV_O_INACTIVE)' failed.
2024-07-16 13:29:55.795+0000: shutting down, reason=crashed
destination hypervisor:
2024-07-16T13:29:54.833540Z qemu-kvm: warning: Deprecated CPU topology (considered invalid): Unsupported clusters parameter mustn't be specified as 1 2024-07-16T13:29:55.579204Z qemu-kvm: Failed to load virtio_pci/modern_queue_state:unused 2024-07-16T13:29:55.579303Z qemu-kvm: Failed to load virtio_pci/modern_state:vqs 2024-07-16T13:29:55.579337Z qemu-kvm: Failed to load virtio/extra_state:extra_state 2024-07-16T13:29:55.579381Z qemu-kvm: Failed to load virtio-rng:virtio 2024-07-16T13:29:55.579414Z qemu-kvm: error while loading state for instance 0x0 of device '0000:00:02.5:00.0/virtio-rng' 2024-07-16T13:29:55.579577Z qemu-kvm: load of migration failed: Input/output error 2024-07-16 13:29:55.790+0000: shutting down, reason=crashed
Please provide the package NVR for which bug is seen:
How reproducible:
100%
Steps to reproduce
- boot an instance with local storage.
- pause the instance
- live migrate the instance from edpm-compute-0 to edpm-compute-1 with block migration
- live migrate the instance back from edpm-compute-1 to edpm-compute-0 with block migration
Expected results
The second migration also succeeds.
Actual results
The second migration fails due to qemu-crash on the the source hypervisor of the second migration.
- causes
-
-
- Backlog
-
- is blocked by
-
RHEL-54296 Provide QMP command for block device reactivation after migration [rhel-9.5]
-
- Planning
-
-
-
- Planning
-
- links to
