# August 2025 Release The latest release of space.vars.company\_name space.vars.PRODUCT\_NAME includes new features, usability improvements, and resolved issues to enhance product stability and performance. ### New Features **Networking Service Upgrade to 2024.1 (Caracal)** Upgraded the core components of the networking service to the 2024.1 (Caracal) release. * Metadata over IPv6 is now supported in the OVN driver. **Storage Service Upgrade to 2025.1 (Epoxy)** Upgraded the core components of the storage service to the 2025.1 (Epoxy) release. * New features and bug fixes for supported drivers, including NetApp, HPE 3PAR, HPE Alletra, and Pure Storage. * Improvements to multipath setup and management. * The NetApp ONTAP driver now supports Certificate-Based-Authentication (CBA) for operators that desire certificate-based authentication instead of user and password. Note: The options for cert-auth take precedence; if all the auth options are defined in the config (both cert and legacy), the legacy ones are ignored. **Tintri Storage Driver Support** Added Tintri driver as a storage option, eliminating manual deployment workflows. UI provides Tintri backend configuration through standard persistent-storage controls, streamlining enterprise storage deployment for production workloads. **Ubuntu 24.04 Support** Ubuntu 24.04 is now supported as a hypervisor host. Support for self-hosted management nodes is coming soon. ### Enhancements **Streamlined GPU Configuration with Automated vGPU Host Setup** The GPU configuration is now simplified by automating the vGPU host setup process that previously required manual script execution. The system now handles vGPU profile configuration automatically through the UI and backend processes, eliminating the need for the separate `vGPU host configure` option in the GPU helper script. This enhancement reduces configuration complexity and the risk of user error while accelerating GPU deployment workflows. **Support for Multiple GPU Models** Platform9 supports multiple GPU models per region while maintaining a single GPU model per host configuration across virtualized and Kubernetes workloads. **Enhanced Server Group Management with Direct VM Assignment Controls** In addition to the existing VM management controls, space.vars.PRODUCT\_ACRONYM now supports **Add Server Group** and **Remove Server Group** actions on existing VMs on the VM list and details page. This action is only supported for `Active` and `Stopped` VMs. Supported Operations: * Add: You can add a single VM to a server group at a time, provided it satisfies the existing affinity or anti-affinity policy rules. * Remove: You can remove one or more VMs from their respective server groups. **Soft Affinity Policy Support for Server Groups** Server groups now support both soft affinity and soft anti-affinity policies, in addition to existing hard constraints. UI and API enable the explicit selection of soft or hard policies during server group creation. **Server Group Quota Management Controls** You can now configure server group and member quotas through **Manage Quotas > Compute Quotas**. **Support for Retype Operation on In-Use Volumes** Volume retype supports in-use and attached volumes with driver validation and user confirmation. space.vars.PRODUCT\_ACRONYM provides compatibility warnings and backend support guidance through enhanced UI controls. **Allow Metrics Scraping from an External Source** You can now scrape host Prometheus metrics from an external source using `host-IP:9388`. **Cold Migration and Resize for vTPM-Enabled VMs** You can now cold migrate or resize vTPM-enabled VMs. Live migration for vTPM-enabled VMs is coming soon. **Kubernetes 1.33 Platform Support** space.vars.company\_name supports Kubernetes 1.33 across all managed clusters starting with this release. Enhanced security controls and performance optimizations are deployed alongside the latest container orchestration capabilities. **OCI Registry Support for Airgapped Helm Deployments** You now have support for OCI registries as Helm repositories, enabling airgapped deployments. Please note that this applies only to Kubernetes with space.vars.PRODUCT\_NAME , not to the virtualization features. ### Upgrade Notes The August 2025 release includes improvements to how cluster hosts are managed. Make sure the following configuration is applied to any hosts that may be missing it. **OS Upgrade from Ubuntu 22.04 to 24.04** For all hosts with a hypervisor role, copy over the `apparmor` config from the old location to the new location, as shown below, after upgrading to Ubuntu 24.04. ```bash cp _etc_apparmor.d_abstractions_libvirt-qemu.dpkg-dist _etc_apparmor.d_abstractions_libvirt-qemu systemctl restart libvirtd.service qemu-kvm.service ``` ### Bug Fixes #### Infrastructure Management * Fixed Resolved hypervisor role authorization errors caused by greenlet module version discrepancy. * Fixed Grafana password reset functionality. Users can now reset passwords using their registered email address with Platform9. * Fixed "Identity provider already exists" error when re-enabling the enterprise SSO feature. #### Self-Hosted PCD * Added pre-deployment NTP checks for self-hosted space.vars.PRODUCT\_ACRONYM. #### Compute and Image Services * Fixed and improved hotplug VM memory allocation transparency. Memory reporting now accounts for hotplug reservation overhead, eliminating discrepancies between configured and usable memory. * Fixed Lease configuration now correctly validates expiry dates, ensuring proper VM termination only for VMs with future expiry dates. * Fixed an issue where vTPM VMs would end up in an error state after a host reboot due to auto-start. Now, vTPM VMs will remain in the shutoff state until powered on by the user upon host reboot. #### Storage Service * Fixed the VM cloning workflow in the UI to correctly use the original volume size, eliminating unexpected storage allocation errors. #### Networking Service Added support for jumbo frames in space.vars.PRODUCT\_ACRONYM. Networks created via API or CLI now have the following default MTUs set if not specified: * `8950` for VXLAN * `8942` for GENEVE * `9000` for VLAN & flat networks The MTUs for networks created with the space.vars.PRODUCT\_ACRONYM UI will still use the existing defaults if not specified: * `1450` for VXLAN * `1440` for GENEVE * `1500` for VLAN & flat networks This change is being made to give users the flexibility to set MTUs larger than 1500 through either CLI or UI if needed. #### Kubernetes on space.vars.PRODUCT\_NAME * Resolved SSO login failures when accessing Kubernetes clusters, eliminating “Internal Errors” you may have encountered. * Subnet assignment in multi-subnet networks during cluster provisioning. Clusters now deploy successfully when multiple subnets are available, eliminating the need for arbitrary subnet selection that previously caused provisioning failures. * Tenant isolation issue where clusters appeared under the incorrect tenant. Cluster visibility now filters to the selected tenant, ensuring accurate segregation of multi-tenant resources. * UI crashes in **Kubernetes > Access Control > Roles** when creating roles with "core" API group selection. Role creation workflow now executes successfully, ensuring reliable RBAC configuration. * MIG partition configuration persistence during Kubernetes cluster upgrades. GPU nodes now maintain MIG partitioning settings when upgrading from 1.31 to 1.32, preventing automatic reversion to passthrough mode. This ensures consistent GPU resource allocation across cluster upgrade operations. * An issue occurred where worker nodes in **Infrastructure > Clusters** incorrectly displayed `Running` machine status even when the underlying virtual machines had stopped or failed. Machine status indicators now accurately reflect the true state of worker node infrastructure, reducing troubleshooting time when monitoring cluster health. * The Delete button for Kubernetes clusters continued to be in the `Deleting` state. The delete button now disables when clusters are marked for deletion, preventing redundant operations. * Rebalancing frequency now shows minute units in the **Edit** **Cluster** workflow UI. * The **Upgrade** button is disabled for active upgrades, preventing multiple operations through improved status tracking. * BYOH functionality now works correctly after resolving missing tenant labels in deployment configurations. * The Cluster dropdown filter now functions correctly across the Config Maps Details, Secrets Details, and Custom Resource Definitions pages. Prevents page breaks when switching between clusters in these pages. * Byohctl now defaults to 'service' when the parameter is omitted. Removed required designation, simplifying usage. * Byocluster listings now display `Ready` status column with standard cluster metadata. #### Miscellaneous * Fixed the inconsistent custom theme application. Theme colors now apply consistently across all regions. ### Known Limitations * VM Migration During Ubuntu Host Upgrades: VM migration is only supported from hosts running Ubuntu 22.04 to Ubuntu 24.04, but not vice versa. To ensure a successful upgrade, you must disable VMHA and DRR features and drain each host before proceeding with the Ubuntu version upgrade. * Rescue mode is only supported for VMs with ephemeral storage. The rescue operation does not work for instances backed by volumes. Users attempting to rescue a volume-backed instance will encounter failures. * No support exists for override configuration for the Image Library host(s). Any manual changes made to the configuration file will not persist across upgrades. * Mors Pod Stability with Leases: When leases are implemented, the mors pod may become unresponsive, preventing the timely execution of operations such as VM delete/stop. As a workaround, the mors pod may need to be restarted manually to restore responsiveness. A permanent fix is planned for a future release. ### Known Issues * GPU Passthrough is currently supported on host kernel versions up to 6.5. * For vGPU support, please refer to the [GPU documentation](https://platform9.com/docs/private-cloud-director/2025.10/private-cloud-director/troubleshooting-gpu-support) for more information on the issues below: * If a GPU PCI device is already bound to a driver/module, it needs to be unbound to enable vGPU on the same PCI. * If a GPU host running vGPU VMs is rebooted, vGPU VMs aren't recovered automatically. * When you assign multiple storage backends to a host and then remove and re-add them, you may need to manually re-enable them. 1. Find the backends to enable by running the following to list services: ```bash pcdctl volume service list ``` 2. Re-enable the required backend(s) using the following command format: ```bash pcdctl volume service set --enable @ cinder-volume ``` * VM HA and DRR do not support vTPM-enabled VMs. Live migration and evacuation are not possible, so these VMs will not be migrated automatically. * VM HA and DRR do not work for hot-plug VMs when using zero-disk flavors. Live migration and evacuation are not possible, so these VMs will not be migrated automatically. * VM HA does not support the evacuation of VMs belonging to server groups with a hard affinity policy, so these VMs will not be evacuated automatically in case of host failure. * If a host with a persistent storage role is assigned and goes down, and VMs running on that host are also served their block storage volumes from the same host, there is a known race condition that may cause the evacuation of those VMs to fail. To avoid this, we recommend assigning a block storage role to hosts that are not assigned a hypervisor role. This issue is being fixed in the December release of Private Cloud Director. * If you are using NFS as the backend for block storage, set the `image_volume_cache_enabled` flag to `false`. If the flag is set to `true`, creating a VM from a cached image volume may lead to incorrect root disk sizing. * `pcdctl config set` command is not supported for a user with MFA enabled. * Image upload to encrypted volumes is currently unsupported. Volume encryption only works with empty volumes at this time. * SSO users cannot log in to PCD Grafana. * If you have a network with a DNS domain assigned, and one of its subnets has DNS Publish Fixed IP enabled, then a port created on any subnet within that network will publish a DNS record, irrespective of the subnet's DNS Publish Fixed IP setting. * When the control plane is upgraded, but hosts are not, VM traffic may be impacted. Mixed-version environments may not function as expected due to the upgrade from OVN to Caracal. When the ovn-controller package is reinstalled and the OVN controller process restarts, a brief disruption of VM traffic, including transient packet drops, may be observed. Traffic recovers automatically once the service is back up and flows are reprogrammed. It is recommended to complete all host upgrades before validating workloads or testing network connectivity. --- # Agent Instructions: Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter: ``` GET https://docs.platform9.com/release-notes/august-2025-release.md?ask= ``` The question should be specific, self-contained, and written in natural language. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.