-
Bug
-
Resolution: Unresolved
-
Undefined
-
None
-
CNV v4.20.3
-
None
-
Quality / Stability / Reliability
-
0.42
-
False
-
-
False
-
None
-
-
Important
-
None
Description of problem:
Multiqueue should be off by default, but its ON. It always matches the number of vCPUs. This knob does nothing by default $ oc explain vm.spec.template.spec.domain.devices.blockMultiQueue GROUP: kubevirt.io KIND: VirtualMachine VERSION: v1 FIELD: blockMultiQueue <boolean> DESCRIPTION: Whether or not to enable virtio multi-queue for block devices. Defaults to false. <-----------------
Version-Release number of selected component (if applicable):
4.20.3
How reproducible:
100%
Steps to Reproduce:
On a VM with >=2 vCPUs (example below has 10 vCPUs): # oc get vm windows-2022-postgres -o yaml | yq '.spec.template.spec.domain.devices.blockMultiQueue' false # oc rsh virt-launcher-windows-2022-postgres-6s257 $ virsh qemu-monitor-command 1 --hmp "info qtree" | grep queues num-queues = 10 (0xa). <--- virtio-blk num_queues = 10 (0xa) <--- virtio-scsi
Actual results:
multiqueue is always enabled, unable to disable it in VM spec.
Expected results:
multiqueue is disabled if blockMultiQueue: false
Additional info:
In the XML add queues='1' on the driver to disable multiqueue, the default on RHEL side is to enable it and match the vCPUs. Doing nothing on the XML doesn't disable it.
- is related to
-
RHEL-141510 "MEMORY MANAGEMENT" BSOD with no storage delay and only on virtio-blk
-
- Release Pending
-
- links to