From a4145af7d52dc460981d73b71a19bf536e0fe3e9 Mon Sep 17 00:00:00 2001 From: Abigail McCarthy <20771501+a-mccarthy@users.noreply.github.com> Date: Tue, 3 Feb 2026 15:11:06 -0500 Subject: [PATCH 1/4] Add openshift clusterpolicy upgrade parameters Signed-off-by: Abigail McCarthy <20771501+a-mccarthy@users.noreply.github.com> --- gpu-operator/gpu-driver-upgrades.rst | 19 ++++++++++++++++++- 1 file changed, 18 insertions(+), 1 deletion(-) diff --git a/gpu-operator/gpu-driver-upgrades.rst b/gpu-operator/gpu-driver-upgrades.rst index 788962b41..233859667 100644 --- a/gpu-operator/gpu-driver-upgrades.rst +++ b/gpu-operator/gpu-driver-upgrades.rst @@ -64,6 +64,15 @@ The controller automates the upgrade process and generates metrics and events so --type='json' \ -p='[{"op": "replace", "path": "/spec/driver/version", "value":"510.85.02"}]' + + If you are using Openshift, you must update the ``driver.version``, ``driver.repository`` and ``driver.image`` values in the cluster policy. + + .. code-block:: console + + $ kubectl patch clusterpolicies.nvidia.com/cluster-policy \ + --type='json' \ + -p='[{"op": "replace", "path": "/spec/driver/version", "value":"510.85.02"},{"op": "replace", "path": "/spec/driver/repository", "value":"nvcr.io/nvidia"},{"op": "replace", "path": "/spec/driver/image", "value":"driver"}]' + 2. (Optional) For each node, monitor the upgrade status: .. code-block:: console @@ -287,7 +296,15 @@ In addition, no new features will be added to the ``k8s-driver-manager`` moving .. code-block:: console - $ kubectl patch clusterpolicies.nvidia.com/cluster-policy --type='json' -p='[{"op": "replace", "path": "/spec/driver/version", "value":"510.85.02"}]' + $ kubectl patch clusterpolicies.nvidia.com/cluster-policy \ + --type='json' \ + -p='[{"op": "replace", "path": "/spec/driver/version", "value":"510.85.02"}, \ + {"op": "replace", "path": "/spec/driver/repository", "value":"nvcr.io/nvidia"}, \ + {"op": "replace", "path": "/spec/driver/image", "value":"driver"}]' + + .. note:: + + If you are using OpenShift, you must include the ``driver.repository`` and ``driver.image`` fields as shown in the command above. 2. (Optional) To monitor the status of the upgrade, watch the deployment of the new driver pod on GPU worker nodes: From cbe95f48c88c3403de859b62c927f9e0c51f126b Mon Sep 17 00:00:00 2001 From: Abigail McCarthy <20771501+a-mccarthy@users.noreply.github.com> Date: Tue, 3 Feb 2026 15:26:09 -0500 Subject: [PATCH 2/4] Update driver version Signed-off-by: Abigail McCarthy <20771501+a-mccarthy@users.noreply.github.com> --- gpu-operator/gpu-driver-upgrades.rst | 16 +++++----------- 1 file changed, 5 insertions(+), 11 deletions(-) diff --git a/gpu-operator/gpu-driver-upgrades.rst b/gpu-operator/gpu-driver-upgrades.rst index 233859667..50c6e8b49 100644 --- a/gpu-operator/gpu-driver-upgrades.rst +++ b/gpu-operator/gpu-driver-upgrades.rst @@ -62,7 +62,7 @@ The controller automates the upgrade process and generates metrics and events so $ kubectl patch clusterpolicies.nvidia.com/cluster-policy \ --type='json' \ - -p='[{"op": "replace", "path": "/spec/driver/version", "value":"510.85.02"}]' + -p='[{"op": "replace", "path": "/spec/driver/version", "value":"580.95.05"}]' If you are using Openshift, you must update the ``driver.version``, ``driver.repository`` and ``driver.image`` values in the cluster policy. @@ -71,7 +71,9 @@ The controller automates the upgrade process and generates metrics and events so $ kubectl patch clusterpolicies.nvidia.com/cluster-policy \ --type='json' \ - -p='[{"op": "replace", "path": "/spec/driver/version", "value":"510.85.02"},{"op": "replace", "path": "/spec/driver/repository", "value":"nvcr.io/nvidia"},{"op": "replace", "path": "/spec/driver/image", "value":"driver"}]' + -p='[{"op": "replace", "path": "/spec/driver/version", "value":"580.95.05"}, + {"op": "replace", "path": "/spec/driver/repository", "value":"nvcr.io/nvidia"}, + {"op": "replace", "path": "/spec/driver/image", "value":"driver"}]' 2. (Optional) For each node, monitor the upgrade status: @@ -296,15 +298,7 @@ In addition, no new features will be added to the ``k8s-driver-manager`` moving .. code-block:: console - $ kubectl patch clusterpolicies.nvidia.com/cluster-policy \ - --type='json' \ - -p='[{"op": "replace", "path": "/spec/driver/version", "value":"510.85.02"}, \ - {"op": "replace", "path": "/spec/driver/repository", "value":"nvcr.io/nvidia"}, \ - {"op": "replace", "path": "/spec/driver/image", "value":"driver"}]' - - .. note:: - - If you are using OpenShift, you must include the ``driver.repository`` and ``driver.image`` fields as shown in the command above. + $ kubectl patch clusterpolicies.nvidia.com/cluster-policy --type='json' -p='[{"op": "replace", "path": "/spec/driver/version", "value":"580.95.05"}]' 2. (Optional) To monitor the status of the upgrade, watch the deployment of the new driver pod on GPU worker nodes: From 62f62a781401288b8be4f222f300fa03a854bfdc Mon Sep 17 00:00:00 2001 From: Abigail McCarthy <20771501+a-mccarthy@users.noreply.github.com> Date: Tue, 3 Feb 2026 15:38:28 -0500 Subject: [PATCH 3/4] Fix highlighting Signed-off-by: Abigail McCarthy <20771501+a-mccarthy@users.noreply.github.com> --- gpu-operator/gpu-driver-upgrades.rst | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) diff --git a/gpu-operator/gpu-driver-upgrades.rst b/gpu-operator/gpu-driver-upgrades.rst index 50c6e8b49..9f60b2648 100644 --- a/gpu-operator/gpu-driver-upgrades.rst +++ b/gpu-operator/gpu-driver-upgrades.rst @@ -71,9 +71,7 @@ The controller automates the upgrade process and generates metrics and events so $ kubectl patch clusterpolicies.nvidia.com/cluster-policy \ --type='json' \ - -p='[{"op": "replace", "path": "/spec/driver/version", "value":"580.95.05"}, - {"op": "replace", "path": "/spec/driver/repository", "value":"nvcr.io/nvidia"}, - {"op": "replace", "path": "/spec/driver/image", "value":"driver"}]' + -p='[{"op": "replace", "path": "/spec/driver/version", "value":"580.95.05"},{"op": "replace", "path": "/spec/driver/repository", "value":"nvcr.io/nvidia"},{"op": "replace", "path": "/spec/driver/image", "value":"driver"}]' 2. (Optional) For each node, monitor the upgrade status: From ecdb11ce1e3ace799305c943e0de70ce5a7d7945 Mon Sep 17 00:00:00 2001 From: Abigail McCarthy <20771501+a-mccarthy@users.noreply.github.com> Date: Wed, 4 Feb 2026 12:59:21 -0500 Subject: [PATCH 4/4] include extra params to upgrades without controller Signed-off-by: Abigail McCarthy <20771501+a-mccarthy@users.noreply.github.com> --- gpu-operator/gpu-driver-upgrades.rst | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/gpu-operator/gpu-driver-upgrades.rst b/gpu-operator/gpu-driver-upgrades.rst index 9f60b2648..570e7aaa8 100644 --- a/gpu-operator/gpu-driver-upgrades.rst +++ b/gpu-operator/gpu-driver-upgrades.rst @@ -296,7 +296,7 @@ In addition, no new features will be added to the ``k8s-driver-manager`` moving .. code-block:: console - $ kubectl patch clusterpolicies.nvidia.com/cluster-policy --type='json' -p='[{"op": "replace", "path": "/spec/driver/version", "value":"580.95.05"}]' + $ kubectl patch clusterpolicies.nvidia.com/cluster-policy --type='json' -p='[{"op": "replace", "path": "/spec/driver/version", "value":"580.95.05"},{"op": "replace", "path": "/spec/driver/repository", "value":"nvcr.io/nvidia"},{"op": "replace", "path": "/spec/driver/image", "value":"driver"}]' 2. (Optional) To monitor the status of the upgrade, watch the deployment of the new driver pod on GPU worker nodes: