Bug 2117186

I had a similar issue but with the kubevirt_migrate_vmi_succeeded_total metric, following are the steps for reproduction:

1. kubectl apply examples/vmi-migratable.yaml
2. kubectl apply examples/migration-job.yaml
3. kubectl exec -it /bin/bash <source_virt_handler_pod_name> -n {kubevirt, openshift-cnv}
4. (From inside the virt-handler pod) curl -kL https://locahost:8443/metrics | grep kubevirt_migrate_vmi_succeeded_total

Please update 'target release' according to your plans.

Even this bug is related to metrics, I change the component to Infra to keep it on the radar of the infra team.

Restoring the correct state of the bug.

*** Bug 2187304 has been marked as a duplicate of this bug. ***

There is a great analysis at https://bugzilla.redhat.com/show_bug.cgi?id=2187304#c0 .

@dholler can you please summarize in a sentence how the analysis relates to this and means for this bug?

The bug is that kubevirt does not get the related values from libvirt.
https://bugzilla.redhat.com/show_bug.cgi?id=2187304#c0  points out as one possible reason, that kubevirt is not using the libvirt API like expected.
Another possible reason is, that libvirt itself has a bug.

Thanks, Dominik.

Moving this over to virtualization.

From https://listman.redhat.com/archives/libvir-list/2023-September/241935.html

"This data was available with the old-style disk migration where disk
data was migrated over the migration stream.

As that was deprecated by qemu, libvirt switched to a new approach where
we migrate the disks by using NBD protocol and copying them via
'blockdev-mirror'.

Now the qemu blockjobs themselves do not report the actual bandwidth used,
just the progress. Since libvirt doesn't want to pointlessly poll this
data just to calculate bandwidth, the data is no longer available."

We are not migrating the disk, since it is supposed to be saved to persistent storage. From what I know we are copying only the memory.
Could it be that we only need the kubevirt_migrate_vmi_memory_transfer_rate_bytes?

(In reply to Shirly Radco from comment #17)
> We are not migrating the disk, since it is supposed to be saved to
> persistent storage. From what I know we are copying only the memory.
> Could it be that we only need the
> kubevirt_migrate_vmi_memory_transfer_rate_bytes?

If I create and migrate a VM with the following config:

---
apiVersion: kubevirt.io/v1
kind: VirtualMachineInstance
metadata:
  labels:
    special: vmi-nocloud
  name: vmi-nocloud
spec:
  networks:
    - name: default
      pod: { }
  domain:
    devices:
      interfaces:
        - masquerade: { }
          name: default
      disks:
      - disk:
          bus: virtio
        name: containerdisk
      - disk:
          bus: virtio
        name: cloudinitdisk
      - disk:
          bus: virtio
        name: emptydisk
    resources:
      requests:
        memory: 128Mi
  terminationGracePeriodSeconds: 0
  volumes:
  - containerDisk:
      image: registry:5000/kubevirt/cirros-container-disk-demo:devel
    name: containerdisk
  - cloudInitNoCloud:
      userData: |
        #!/bin/sh

        echo 'printed from cloud-init userdata'
    name: cloudinitdisk
  - emptyDisk:
      capacity: 2Gi
    name: emptydisk

libvirt reports disk being migrated alongside memory

Joao, correct.
containerDisks do include block migration