KubeVirt
v1.8.2Orchestration & ManagementKubeVirt v1.8.2 is a patch release with 12 bug fixes covering hotplug volumes, live migration, TLS config, qcow2 disk handling, and multi-arch VM stability.
breakingBackend storage volume naming changed — check for impacts
VMs using backend storage volumes will now report the volume name as 'persistent-state-for-this-vm' instead of a name derived from the VM name. If you have monitoring alerts, scripts, or tooling that match on the old volume name pattern, they will silently break. Audit any automation that references backend storage volume names before upgrading.
enhancementUpgrade s390x and ARM64 clusters — arch-specific VM failures resolved
Two separate fixes land here: ARM64 guests now correctly expose SMBIOS system info, and s390x VMs failing to create due to unsupported PCIe root-port controllers from the v3 PCI topology change are fixed. If you're running multi-arch clusters and held back on v1.8.x due to these issues, v1.8.2 is the version to move to.
enhancementTLS config from KubeVirt CR now respected by sync-controller and virt-exportserver
Previously, the sync-controller healthz server and virt-exportserver ignored the TLSConfiguration set in the KubeVirt CR, potentially using weaker defaults. After upgrading, verify your KubeVirt CR TLS settings are correct — they'll actually take effect now. This matters most for clusters with strict cipher or TLS version requirements.
主な変更 (6)
- virt-handler now auto-restarts the notify server when domain-notify.sock is deleted, preventing silent VM communication failures
- Hotplug volumes stuck in 'Detaching' phase are now correctly resolved
- Fixed migration status not reporting 'succeeded' after decentralized live migration with compute migration
- qcow2 overlay disks now correctly resolve source, preventing wrong disk expansion and incorrect cache/IO mode
- s390x VM creation failures from unsupported PCIe root-port controllers (introduced in v3 PCI topology) are fixed
- Backend storage volumes now use a stable name 'persistent-state-for-this-vm' instead of embedding the VM name