From 59d4f65480f3779c9a2f35ae37a3da257fd120c5 Mon Sep 17 00:00:00 2001
From: Sheogorath <sheogorath@shivering-isles.com>
Date: Sun, 26 Mar 2023 21:40:33 +0200
Subject: [PATCH] fix(drivers): Downgrade amd-gpu driver deployment to version
1.18.0
Due to a currently unsolved issue, the device plugin reports 0 available
AMD GPUs on system that have fully functional GPUs ready to go. This
change in behaviour appeared when upgrading from Kubernetes 1.24.11 to
1.25.8 and is currently unclear.
References:
https://github.com/RadeonOpenCompute/k8s-device-plugin/issues/31
---
infrastructure/drivers/amd-gpu.yaml | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/infrastructure/drivers/amd-gpu.yaml b/infrastructure/drivers/amd-gpu.yaml
index 000f78a8c..09c6e7216 100644
--- a/infrastructure/drivers/amd-gpu.yaml
+++ b/infrastructure/drivers/amd-gpu.yaml
@@ -25,6 +25,11 @@ spec:
version: 0.6.0
interval: 5m
values:
+ nfd:
+ enabled: true
+ dp:
+ image:
+ tag: 1.18.0
namespace: drivers-system
securityContext:
allowPrivilegeEscalation: false
--
GitLab