Skip to content
Closed
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 10 additions & 2 deletions gpu-operator/release-notes.rst
Original file line number Diff line number Diff line change
Expand Up @@ -120,8 +120,6 @@ Known Issues
These defaults prevent a disruptive driver upgrade.
However, the following deployments need NVIDIA Driver Manager for Kubernetes v0.13.0:

- Red Hat OpenShift Container Platform in an air-gapped environment, to obtain the fix for uncompressed image layers that can prevent mirroring to Red Hat Quay with ``oc-mirror``.
Refer to `Issue #3166 <https://github.com/NVIDIA/gpu-operator/issues/3166>`__.
- Clusters that use the DRA Driver for NVIDIA GPUs.
- Nodes configured for GPU passthrough with ``vfio-pci``.

Expand Down Expand Up @@ -149,6 +147,16 @@ Known Issues
Ampere GPUs cannot change MIG mode without a GPU reset, so the DRA driver does not publish the full GPU that these pods request.
Refer to `Issue #1580 <https://github.com/kubernetes-sigs/dra-driver-nvidia-gpu/issues/1580>`__.

* For air-gapped or disconnected deployments, use Red Hat Quay 3.17.3 or later to mirror GPU Operator images with ``oc-mirror``.
Quay 3.17.3 successfully mirrors the other tested images; older Quay versions can fail to mirror additional images.
The ``dra-driver-nvidia-gpu:v0.5.1`` image, retained in the default OLM bundle, fails to mirror even with Quay 3.17.3.
No fixed DRA driver image is available.

If you do not use the DRA driver, add its original digest-pinned image reference from the bundle's ``relatedImages`` to ``mirror.blockedImages`` in your ``ImageSetConfiguration``.
If you use the DRA driver, no workaround is currently available for this mirroring failure.

Refer to `Issue #3166 <https://github.com/NVIDIA/gpu-operator/issues/3166>`__.

Post-Release Documentation Updates
----------------------------------

Expand Down