Fix the module "logger" missing when running DCGM (#137)

hippogr · Rui Gao · web-flow · commit c977d35fed5b · 2026-01-16T08:14:55.000+08:00
* add missing module when updating DCGM higher version

* install dcgm with specified version

---------

Co-authored-by: Rui Gao &lt;ruigao@microsoft.com&gt;
diff --git a/src/job-exporter/build/update-dcgm.py b/src/job-exporter/build/update-dcgm.py
@@ -5,6 +5,7 @@
 
 #!/usr/bin/env python3
 
+DCGM_TARGET_VERSION = "1:4.4.1-1"
 
 def get_dcgm_version():
     try:
@@ -72,8 +73,10 @@ def install_latest_dcgm():
         )
         subprocess.run(
             [
-                "apt-get", "install", "--yes", "--install-recommends",
-                "datacenter-gpu-manager-4-cuda12"
+                "apt-get", "install", "--yes",
+                f"datacenter-gpu-manager-4-cuda12={DCGM_TARGET_VERSION}",
+                f"datacenter-gpu-manager-4-core={DCGM_TARGET_VERSION}",
+                f"datacenter-gpu-manager-4-proprietary-cuda12={DCGM_TARGET_VERSION}"
             ],
             check=True
         )

Original file line number	Diff line number	Diff line change
`@@ -5,6 +5,7 @@`
`5`	`5`
`6`	`6`	`#!/usr/bin/env python3`
`7`	`7`
	`8`	`+DCGM_TARGET_VERSION = "1:4.4.1-1"`
`8`	`9`
`9`	`10`	`def get_dcgm_version():`
`10`	`11`	`try:`
`@@ -72,8 +73,10 @@ def install_latest_dcgm():`
`72`	`73`	`)`
`73`	`74`	`subprocess.run(`
`74`	`75`	`[`
`75`		`- "apt-get", "install", "--yes", "--install-recommends",`
`76`		`- "datacenter-gpu-manager-4-cuda12"`
	`76`	`+ "apt-get", "install", "--yes",`
	`77`	`+ f"datacenter-gpu-manager-4-cuda12={DCGM_TARGET_VERSION}",`
	`78`	`+ f"datacenter-gpu-manager-4-core={DCGM_TARGET_VERSION}",`
	`79`	`+ f"datacenter-gpu-manager-4-proprietary-cuda12={DCGM_TARGET_VERSION}"`
`77`	`80`	`],`
`78`	`81`	`check=True`
`79`	`82`	`)`