From 957b8255cd396c2a310f15ea35a8847fbb94324c Mon Sep 17 00:00:00 2001 From: Justin Riley Date: Thu, 21 Nov 2024 09:21:39 -0500 Subject: [PATCH] disable automatic NVIDIA driver upgrades Automatic upgrades of the NVIDIA driver are dangerous in production given that they often happen unattended and are likely to cause issues on at least a subset of nodes. In the best case, user workloads are impacted at unpredictable times. This patch disables automatic upgrades by default. Systems that want automatic upgrades should patch the GPU cluster policy in an overlay. --- nvidia-gpu-operator/base/gpu-cluster-policy.yaml | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/nvidia-gpu-operator/base/gpu-cluster-policy.yaml b/nvidia-gpu-operator/base/gpu-cluster-policy.yaml index d37c6d79..9672c996 100644 --- a/nvidia-gpu-operator/base/gpu-cluster-policy.yaml +++ b/nvidia-gpu-operator/base/gpu-cluster-policy.yaml @@ -38,7 +38,7 @@ spec: repoConfig: configMapName: "" upgradePolicy: - autoUpgrade: true + autoUpgrade: false drain: deleteEmptyDir: false enable: false