What were you trying to do that didn't work?

Same issue find on

Red Hat Enterprise Linux release 10.0 Beta (Coughlan)
6.10.0-15.el10.x86_64
qemu-kvm-9.0.0-4.el10.x86_64
seabios-bin-1.16.3-4.el10.noarch
edk2-ovmf-20240524-1.el10.noarch
libvirt-10.5.0-1.el10.x86_64

Potentially not limited to, but the scenario is:

• In OpenShift Virtualization 4.16 (RHEL 9.4 based), set 2 Windows 2019 Virtual Machines.
• Present a shared LUN from SAN to each, with reservation turned on

$ oc get vm windows-2019 -o yaml  | yq '.spec.template.spec.domain.devices.disks'
[
...
  {
    "lun": {
      "bus": "scsi",
      "reservation": true
    },
    "name": "iscsi-pv"
  }
]

For this, kubevirt generates the following XML

$ oc rsh virt-launcher-windows-2019-hd4h2 virsh dumpxml 1 --xpath //domain//devices//disk
...
<disk type="block" device="lun">
  <driver name="qemu" type="raw" cache="none" error_policy="stop" io="native" discard="unmap"/>
  <source dev="/dev/iscsi-pv" index="1">
    <reservations managed="no">
      <source type="unix" path="/var/run/kubevirt/daemons/pr/pr-helper.sock" mode="client"/>
    </reservations>
  </source>
  <backingStore/>
  <target dev="sdb" bus="scsi"/>
  <alias name="ua-iscsi-pv"/>
  <address type="drive" controller="0" bus="0" target="0" unit="1"/>
</disk>

Once both VMs are up using this shared LUN, create a Failover Cluster and try to run the Validation Test (make sure the disk is offline on the cluster otherwise it won't try to failover the disk).

When it tries to failover the disk, one of the VMs will crash.

341257 04:55:28.245249 write(2<pipe:[3066370]>, "qemu-kvm: ../hw/scsi/scsi-disk.c:558: void scsi_write_data(SCSIRequest *): Assertion `r->req.aiocb == NULL' failed.\n", 116) = 116 <0.000005>

Another simpler way to reproduce this:

1. setup the cluster in the first VM, bring the disk online
2. on the second VM, try to bring the disk online (without any cluster setup)

Please provide the package NVR for which bug is seen:

qemu-kvm-core-8.2.0-11.el9_4.3.x86_64

How reproducible:

Always

Expected results

VM is up, disk can failover and test passes

Actual results

VMs are crashing, cannot use cluster or failover disk