diff --git a/virt/virt-4-11-release-notes.adoc b/virt/virt-4-11-release-notes.adoc index a18cee47baea..fd340c2437e2 100644 --- a/virt/virt-4-11-release-notes.adoc +++ b/virt/virt-4-11-release-notes.adoc @@ -197,6 +197,8 @@ link:https://access.redhat.com/support/offerings/techpreview[Technology Preview [id="virt-4-11-known-issues"] == Known issues +You cannot run {VirtProductName} on a single-stack IPv6 cluster. (link:https://bugzilla.redhat.com/show_bug.cgi?id=2193267[*BZ#2193267*]) + * In a heterogeneous cluster with different compute nodes, virtual machines that have HyperV Reenlightenment enabled cannot be scheduled on nodes that do not support timestamp-counter scaling (TSC) or have the appropriate TSC frequency. (link:https://bugzilla.redhat.com/show_bug.cgi?id=2151169[*BZ#2151169*]) * When you use two pods with different SELinux contexts, VMs with the `ocs-storagecluster-cephfs` storage class fail to migrate and the VM status changes to `Paused`. This is because both pods try to access the shared `ReadWriteMany` CephFS volume at the same time. (link:https://bugzilla.redhat.com/show_bug.cgi?id=2092271[*BZ#2092271*])