From 371dfc672f05f0bda4fa3441beeb261ea5d3ce02 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Efe=20G=C3=B6kdemir?= Date: Tue, 22 Sep 2026 00:59:07 +0300 Subject: [PATCH] docs: document standalone DRA CRD conflicts MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Signed-off-by: Efe Gökdemir --- gpu-operator/release-notes.rst | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/gpu-operator/release-notes.rst b/gpu-operator/release-notes.rst index 1e6d9f471..efcdb7d43 100644 --- a/gpu-operator/release-notes.rst +++ b/gpu-operator/release-notes.rst @@ -189,6 +189,10 @@ Known Issues To avoid conflicting allocations, submit the workloads serially and wait for device preparation to complete, or dedicate separate nodes to container and VFIO workloads. +* If you install the DRA Driver for NVIDIA GPUs from its own Helm chart alongside GPU Operator, both charts install the ComputeDomain custom resource definitions (CRDs). + GitOps tools that track CRD ownership can report a conflict between the two releases. + Do not install the standalone DRA chart with the ``GPUCluster`` workflow. If you use the standalone chart instead, install GPU Operator with ``--skip-crds`` and apply the ``ClusterPolicy`` and ``NVIDIADriver`` CRDs separately before creating those resources. + Post-Release Documentation Updates ----------------------------------