From b4a4395c44d59a5903974255e84f66368581a363 Mon Sep 17 00:00:00 2001
From: Verdi March <marcverd@amazon.com>
Date: Wed, 6 Mar 2024 15:06:37 +0800
Subject: [PATCH] SMHP: slurm exporter to report gpu metrics

---
 .../base-config/utils/install_slurm_exporter.sh               | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/1.architectures/5.sagemaker-hyperpod/LifecycleScripts/base-config/utils/install_slurm_exporter.sh b/1.architectures/5.sagemaker-hyperpod/LifecycleScripts/base-config/utils/install_slurm_exporter.sh
index 194759fa..c8b206e9 100644
--- a/1.architectures/5.sagemaker-hyperpod/LifecycleScripts/base-config/utils/install_slurm_exporter.sh
+++ b/1.architectures/5.sagemaker-hyperpod/LifecycleScripts/base-config/utils/install_slurm_exporter.sh
@@ -10,7 +10,7 @@ if sudo systemctl is-active --quiet slurmctld; then
         echo "Go is already installed."
     fi
     echo "This was identified as the controller node because Slurmctld is running. Begining SLURM Exporter Installation"
-    git clone -b 0.20 https://github.com/vpenso/prometheus-slurm-exporter.git
+    git clone -b development https://github.com/vpenso/prometheus-slurm-exporter.git
     cd prometheus-slurm-exporter
     sudo make && sudo cp bin/prometheus-slurm-exporter /usr/bin/
     sudo tee /etc/systemd/system/prometheus-slurm-exporter.service > /dev/null <<EOF
@@ -19,7 +19,7 @@ Description=Prometheus SLURM Exporter
 
 [Service]
 Environment=PATH=/opt/slurm/bin:\$PATH
-ExecStart=/usr/bin/prometheus-slurm-exporter
+ExecStart=/usr/bin/prometheus-slurm-exporter -gpus-acct
 Restart=on-failure
 RestartSec=15
 Type=simple