Update to DCGM 3.3.0 and ubi9

Dockerfile

@@ -1,14 +1,14 @@
-FROM nvidia/cuda:12.2.0-base-ubuntu22.04
+FROM nvidia/cuda:12.2.2-base-ubuntu22.04
 
 ARG DCGM_VERSION
 ENV DEBIAN_FRONTEND=noninteractive
 
 RUN apt-get update && apt-get install -y --no-install-recommends \
     gnupg2 curl ca-certificates build-essential && \
-    curl -fsSL https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/7fa2af80.pub | apt-key add - && \
+    curl -fsSL https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2204/x86_64/7fa2af80.pub | apt-key add - && \
     curl -s https://storage.googleapis.com/golang/go1.16.5.linux-amd64.tar.gz| tar -v -C /usr/local -xz && \
-    echo "deb https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64 /" > /etc/apt/sources.list.d/cuda.list && \
-    echo "deb https://developer.download.nvidia.com/compute/machine-learning/repos/ubuntu2004/x86_64 /" > /etc/apt/sources.list.d/nvidia-ml.list && \
+    echo "deb https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2204/x86_64 /" > /etc/apt/sources.list.d/cuda.list && \
+    echo "deb https://developer.download.nvidia.com/compute/machine-learning/repos/ubuntu2204/x86_64 /" > /etc/apt/sources.list.d/nvidia-ml.list && \
     apt-get purge --autoremove -y curl \
     && rm -rf /var/lib/apt/lists/*
 

Makefile

@@ -15,9 +15,9 @@
 MKDIR    ?= mkdir
 REGISTRY ?= nvidia
 
-DCGM_VERSION   := 3.2.5
+DCGM_VERSION   := 3.3.0
 GOLANG_VERSION := 1.20
-VERSION        := 3.1.8
+VERSION        := 3.2.0
 FULL_VERSION   := $(DCGM_VERSION)-$(VERSION)
 OUTPUT         := type=oci,dest=/tmp/dcgm-exporter.tar
 PLATFORMS      := linux/amd64,linux/arm64
@@ -30,7 +30,7 @@ NON_TEST_FILES  += cmd/dcgm-exporter/main.go
 MAIN_TEST_FILES := pkg/dcgmexporter/system_info_test.go
 
 .PHONY: all binary install check-format local
-all: ubuntu22.04
+all: ubuntu22.04 ubi9
 
 binary:
 	cd cmd/dcgm-exporter; go build -ldflags "-X main.BuildVersion=${DCGM_VERSION}-${VERSION}"
@@ -49,7 +49,7 @@ check-format:
 
 push:
 	$(MAKE) ubuntu22.04 OUTPUT=type=registry
-	$(MAKE) ubi8 OUTPUT=type=registry
+	$(MAKE) ubi9 OUTPUT=type=registry
 
 local:
 ifeq ($(shell uname -p),aarch64)
@@ -66,3 +66,13 @@ ubuntu22.04:
 		--build-arg "DCGM_VERSION=$(DCGM_VERSION)" \
 		--tag "$(REGISTRY)/dcgm-exporter:$(FULL_VERSION)-ubuntu22.04" \
 		--file docker/Dockerfile.ubuntu22.04 .
+
+ubi9:
+	$(DOCKERCMD) --pull \
+		--output $(OUTPUT) \
+		--platform $(PLATFORMS) \
+		--build-arg "GOLANG_VERSION=$(GOLANG_VERSION)" \
+		--build-arg "DCGM_VERSION=$(DCGM_VERSION)" \
+		--build-arg "VERSION=$(FULL_VERSION)" \
+		--tag "$(REGISTRY)/dcgm-exporter:$(FULL_VERSION)-ubi9" \
+		--file docker/Dockerfile.ubi9 .

README.md

@@ -10,7 +10,7 @@ Official documentation for DCGM-Exporter can be found on [docs.nvidia.com](https
 
 To gather metrics on a GPU node, simply start the `dcgm-exporter` container:
 ```
-$ docker run -d --gpus all --rm -p 9400:9400 nvcr.io/nvidia/k8s/dcgm-exporter:3.2.5-3.1.8-ubuntu20.04
+$ docker run -d --gpus all --rm -p 9400:9400 nvcr.io/nvidia/k8s/dcgm-exporter:3.3.0-3.2.0-ubuntu22.04
 $ curl localhost:9400/metrics
 # HELP DCGM_FI_DEV_SM_CLOCK SM clock frequency (in MHz).
 # TYPE DCGM_FI_DEV_SM_CLOCK gauge

dcgm-exporter.yaml

@@ -18,23 +18,23 @@ metadata:
   name: "dcgm-exporter"
   labels:
     app.kubernetes.io/name: "dcgm-exporter"
-    app.kubernetes.io/version: "3.1.8"
+    app.kubernetes.io/version: "3.2.0"
 spec:
   updateStrategy:
     type: RollingUpdate
   selector:
     matchLabels:
       app.kubernetes.io/name: "dcgm-exporter"
-      app.kubernetes.io/version: "3.1.8"
+      app.kubernetes.io/version: "3.2.0"
   template:
     metadata:
       labels:
         app.kubernetes.io/name: "dcgm-exporter"
-        app.kubernetes.io/version: "3.1.8"
+        app.kubernetes.io/version: "3.2.0"
       name: "dcgm-exporter"
     spec:
       containers:
-      - image: "nvcr.io/nvidia/k8s/dcgm-exporter:3.2.5-3.1.8-ubuntu20.04"
+      - image: "nvcr.io/nvidia/k8s/dcgm-exporter:3.3.0-3.2.0-ubuntu22.04"
         env:
         - name: "DCGM_EXPORTER_LISTEN"
           value: ":9400"
@@ -64,11 +64,11 @@ metadata:
   name: "dcgm-exporter"
   labels:
     app.kubernetes.io/name: "dcgm-exporter"
-    app.kubernetes.io/version: "3.1.8"
+    app.kubernetes.io/version: "3.2.0"
 spec:
   selector:
     app.kubernetes.io/name: "dcgm-exporter"
-    app.kubernetes.io/version: "3.1.8"
+    app.kubernetes.io/version: "3.2.0"
   ports:
   - name: "metrics"
     port: 9400

deployment/Chart.yaml

@@ -1,9 +1,9 @@
 apiVersion: v2
 name: dcgm-exporter
 description: A Helm chart for DCGM exporter
-version: "3.1.8"
+version: "3.2.0"
 kubeVersion: ">= 1.19.0-0"
-appVersion: "3.1.8"
+appVersion: "3.2.0"
 sources:
 - https://github.com/nvidia/dcgm-exporter
 home: https://github.com/nvidia/dcgm-exporter/

deployment/values.yaml

@@ -17,7 +17,7 @@ image:
   pullPolicy: IfNotPresent
   # Image tag defaults to AppVersion, but you can use the tag key
   # for the image tag, e.g:
-  tag: 3.2.5-3.1.8-ubuntu20.04
+  tag: 3.3.0-3.2.0-ubuntu22.04
 
 # Change the following reference to "/etc/dcgm-exporter/default-counters.csv"
 # to stop profiling metrics from DCGM

docker/Dockerfile.ubi9

@@ -0,0 +1,41 @@
+ARG GOLANG_VERSION
+FROM golang:$GOLANG_VERSION AS builder
+WORKDIR /go/src/github.com/NVIDIA/dcgm-exporter
+
+COPY . .
+
+RUN make binary check-format
+
+FROM nvcr.io/nvidia/cuda:12.2.2-base-ubi9
+LABEL io.k8s.display-name="NVIDIA DCGM Exporter"
+
+ARG DCGM_VERSION
+
+RUN dnf clean expire-cache && dnf install -y datacenter-gpu-manager-${DCGM_VERSION} libcap
+
+COPY --from=builder /go/src/github.com/NVIDIA/dcgm-exporter/cmd/dcgm-exporter/dcgm-exporter /usr/bin/
+COPY etc /etc/dcgm-exporter
+
+ENV NVIDIA_VISIBLE_DEVICES=all
+# disable all constraints on the configurations required by NVIDIA container toolkit
+ENV NVIDIA_DISABLE_REQUIRE="true"
+# Required for DCP metrics
+ENV NVIDIA_DRIVER_CAPABILITIES=compute,utility,compat32
+
+ARG VERSION
+
+LABEL io.k8s.display-name="NVIDIA DCGM Eporter"
+LABEL name="NVIDIA DCGM Exporter"
+LABEL vendor="NVIDIA"
+LABEL version="${VERSION}"
+LABEL release="N/A"
+LABEL summary="Exports GPU Metrics to Prometheus"
+LABEL description="See summary"
+
+COPY ./LICENSE ./licenses/LICENSE
+
+ENV NO_SETCAP=
+COPY docker/dcgm-exporter-entrypoint.sh /usr/local/dcgm/dcgm-exporter-entrypoint.sh
+RUN chmod +x /usr/local/dcgm/dcgm-exporter-entrypoint.sh
+
+ENTRYPOINT ["/usr/local/dcgm/dcgm-exporter-entrypoint.sh"]

go.mod

@@ -28,7 +28,7 @@ replace (
 )
 
 require (
-	github.com/NVIDIA/go-dcgm v0.0.0-20230816170901-d898cc7820fe
+	github.com/NVIDIA/go-dcgm v0.0.0-20231107230254-2e092a7526b3
 	github.com/NVIDIA/gpu-monitoring-tools v0.0.0-20211102125545-5a2c58442e48
 	github.com/gorilla/mux v1.8.0
 	github.com/sirupsen/logrus v1.9.3
@@ -47,7 +47,6 @@ require (
 	github.com/Microsoft/go-winio v0.6.0 // indirect
 	github.com/beorn7/perks v1.0.1 // indirect
 	github.com/bits-and-blooms/bitset v1.8.0 // indirect
-	github.com/blang/semver v3.5.1+incompatible // indirect
 	github.com/blang/semver/v4 v4.0.0 // indirect
 	github.com/cespare/xxhash/v2 v2.2.0 // indirect
 	github.com/cpuguy83/go-md2man/v2 v2.0.2 // indirect
@@ -64,7 +63,6 @@ require (
 	github.com/google/go-cmp v0.5.9 // indirect
 	github.com/google/gofuzz v1.2.0 // indirect
 	github.com/google/uuid v1.3.0 // indirect
-	github.com/googleapis/gnostic v0.5.5 // indirect
 	github.com/josharian/intern v1.0.0 // indirect
 	github.com/json-iterator/go v1.1.12 // indirect
 	github.com/mailru/easyjson v0.7.7 // indirect
@@ -81,7 +79,6 @@ require (
 	github.com/russross/blackfriday/v2 v2.1.0 // indirect
 	github.com/spf13/pflag v1.0.5 // indirect
 	github.com/xrash/smetrics v0.0.0-20201216005158-039620a65673 // indirect
-	golang.org/x/crypto v0.13.0 // indirect
 	golang.org/x/mod v0.10.0 // indirect
 	golang.org/x/net v0.15.0 // indirect
 	golang.org/x/oauth2 v0.12.0 // indirect