Create multi-arch (amd64,arm64) images. (#41)

* Create multi-arch (amd64,arm64) images.

* Create 'local' build target easy 'docker images' integration

Signed-off-by: Douglas Wightman <dwightman@nvidia.com>

Makefile

@@ -12,22 +12,24 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-DOCKER   ?= docker
 MKDIR    ?= mkdir
 REGISTRY ?= nvidia
 
 DCGM_VERSION   := 2.3.2
 GOLANG_VERSION := 1.17
-VERSION        := 2.6.2
+VERSION        := 2.6.3
 FULL_VERSION   := $(DCGM_VERSION)-$(VERSION)
+OUTPUT         := type=oci,dest=/tmp/dcgm-exporter.tar
+PLATFORMS      := linux/amd64,linux/arm64
+DOCKERCMD      := docker buildx build
 
 NON_TEST_FILES  := pkg/dcgmexporter/dcgm.go pkg/dcgmexporter/gpu_collector.go pkg/dcgmexporter/parser.go
 NON_TEST_FILES  += pkg/dcgmexporter/pipeline.go pkg/dcgmexporter/server.go pkg/dcgmexporter/system_info.go
 NON_TEST_FILES  += pkg/dcgmexporter/types.go pkg/dcgmexporter/utils.go pkg/dcgmexporter/kubernetes.go
 NON_TEST_FILES  += cmd/dcgm-exporter/main.go
 MAIN_TEST_FILES := pkg/dcgmexporter/system_info_test.go
 
-.PHONY: all binary install check-format
+.PHONY: all binary install check-format local
 all: ubuntu20.04 ubi8
 
 binary:
@@ -46,18 +48,29 @@ check-format:
 	test $$(gofmt -l cmd | tee /dev/stderr | wc -l) -eq 0
 
 push:
-	$(DOCKER) push "$(REGISTRY)/dcgm-exporter:$(FULL_VERSION)-ubuntu20.04"
-	$(DOCKER) push "$(REGISTRY)/dcgm-exporter:$(FULL_VERSION)-ubi8"
+	$(MAKE) ubuntu20.04 OUTPUT=type=registry
+	$(MAKE) ubi8 OUTPUT=type=registry
+
+local:
+ifeq ($(shell uname -p),aarch64)
+	$(MAKE) PLATFORMS=linux/arm64 OUTPUT=type=docker DOCKERCMD='docker build'
+else
+	$(MAKE) PLATFORMS=linux/amd64 OUTPUT=type=docker DOCKERCMD='docker build'
+endif
 
 ubuntu20.04:
-	$(DOCKER) build --pull \
+	$(DOCKERCMD) --pull \
+		--output $(OUTPUT) \
+		--platform $(PLATFORMS) \
 		--build-arg "GOLANG_VERSION=$(GOLANG_VERSION)" \
 		--build-arg "DCGM_VERSION=$(DCGM_VERSION)" \
 		--tag "$(REGISTRY)/dcgm-exporter:$(FULL_VERSION)-ubuntu20.04" \
 		--file docker/Dockerfile.ubuntu20.04 .
 
 ubi8:
-	$(DOCKER) build --pull \
+	$(DOCKERCMD) --pull \
+		--output $(OUTPUT) \
+		--platform $(PLATFORMS) \
 		--build-arg "GOLANG_VERSION=$(GOLANG_VERSION)" \
 		--build-arg "DCGM_VERSION=$(DCGM_VERSION)" \
 		--build-arg "VERSION=$(FULL_VERSION)" \

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:2.3.2-2.6.2-ubuntu20.04
+$ docker run -d --gpus all --rm -p 9400:9400 nvcr.io/nvidia/k8s/dcgm-exporter:2.3.2-2.6.3-ubuntu20.04
 $ curl localhost:9400/metrics
 # HELP DCGM_FI_DEV_SM_CLOCK SM clock frequency (in MHz).
 # TYPE DCGM_FI_DEV_SM_CLOCK gauge
@@ -135,6 +135,26 @@ You will also find the `json` file on this repo under `grafana/dcgm-exporter-das
 
 Pull requests are accepted!
 
+
+### Building the containers
+
+This project uses [docker buildx](https://docs.docker.com/buildx/working-with-buildx/) for multi-arch image creation. Follow the instructions on that page to get a working builder instance for creating these containers. Some other useful build options follow.
+
+Builds local images based on the machine architecture and makes them available in 'docker images'
+```
+make local
+```
+
+Build the ubuntu image and export to 'docker images'
+```
+make ubuntu20.04 PLATFORMS=linux/amd64 OUTPUT=type=docker
+```
+
+Build and push the images to some other 'private_registry'
+```
+make REGISTRY=<private_registry> push
+```
+
 ## Issues and Contributing
 
 [Checkout the Contributing document!](CONTRIBUTING.md)

dcgm-exporter.yaml

@@ -18,23 +18,23 @@ metadata:
   name: "dcgm-exporter"
   labels:
     app.kubernetes.io/name: "dcgm-exporter"
-    app.kubernetes.io/version: "2.6.2"
+    app.kubernetes.io/version: "2.6.3"
 spec:
   updateStrategy:
     type: RollingUpdate
   selector:
     matchLabels:
       app.kubernetes.io/name: "dcgm-exporter"
-      app.kubernetes.io/version: "2.6.2"
+      app.kubernetes.io/version: "2.6.3"
   template:
     metadata:
       labels:
         app.kubernetes.io/name: "dcgm-exporter"
-        app.kubernetes.io/version: "2.6.2"
+        app.kubernetes.io/version: "2.6.3"
       name: "dcgm-exporter"
     spec:
       containers:
-      - image: "nvcr.io/nvidia/k8s/dcgm-exporter:2.3.2-2.6.2-ubuntu20.04"
+      - image: "nvcr.io/nvidia/k8s/dcgm-exporter:2.3.2-2.6.3-ubuntu20.04"
         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: "2.6.2"
+    app.kubernetes.io/version: "2.6.3"
 spec:
   selector:
     app.kubernetes.io/name: "dcgm-exporter"
-    app.kubernetes.io/version: "2.6.2"
+    app.kubernetes.io/version: "2.6.3"
   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: "2.6.2"
+version: "2.6.3"
 kubeVersion: ">= 1.13.0-0"
-appVersion: "2.6.2"
+appVersion: "2.6.3"
 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: 2.3.2-2.6.2-ubuntu20.04
+  tag: 2.3.2-2.6.3-ubuntu20.04
 
 # Comment the following line to stop profiling metrics from DCGM
 arguments: ["-f", "/etc/dcgm-exporter/dcp-metrics-included.csv"]

docker/Dockerfile.ubi8

@@ -6,13 +6,12 @@ COPY . .
 
 RUN make binary check-format
 
-FROM nvcr.io/nvidia/cuda:11.4.2-base-ubi8
+FROM nvcr.io/nvidia/cuda:11.6.0-base-ubi8
 LABEL io.k8s.display-name="NVIDIA DCGM Exporter"
 
 ARG DCGM_VERSION
-RUN dnf config-manager --add-repo https://developer.download.nvidia.com/compute/cuda/repos/rhel8/x86_64/cuda-rhel8.repo &&  \
-    dnf clean expire-cache
-RUN dnf install -y datacenter-gpu-manager-${DCGM_VERSION}
+
+RUN dnf clean expire-cache && dnf install -y datacenter-gpu-manager-${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

docker/Dockerfile.ubuntu20.04

@@ -6,23 +6,15 @@ COPY . .
 
 RUN make binary check-format
 
-FROM nvcr.io/nvidia/cuda:11.4.2-base-ubuntu20.04
+FROM nvcr.io/nvidia/cuda:11.6.0-base-ubuntu20.04
 LABEL io.k8s.display-name="NVIDIA DCGM Exporter"
 
 COPY --from=builder /go/src/github.com/NVIDIA/dcgm-exporter/cmd/dcgm-exporter/dcgm-exporter /usr/bin/
 COPY etc /etc/dcgm-exporter
 
 ARG DCGM_VERSION
 RUN apt-get update && apt-get install -y --no-install-recommends \
-    libcap2-bin gnupg2 curl ca-certificates && \
-    curl -fsSL https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/7fa2af80.pub | apt-key add - && \
-    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 && \
-    apt-get purge --autoremove -y curl \
-    && rm -rf /var/lib/apt/lists/*
-
-RUN apt-get update && apt-get install -y --no-install-recommends \
-    datacenter-gpu-manager=1:${DCGM_VERSION} && apt-get purge --autoremove -y openssl
+    datacenter-gpu-manager=1:${DCGM_VERSION} libcap2-bin && apt-get purge --autoremove -y openssl
 
 # Required for DCP metrics
 ENV NVIDIA_DRIVER_CAPABILITIES=compute,utility,compat32

service-monitor.yaml

@@ -18,12 +18,12 @@ metadata:
   name: "dcgm-exporter"
   labels:
     app.kubernetes.io/name: "dcgm-exporter"
-    app.kubernetes.io/version: "2.6.2"
+    app.kubernetes.io/version: "2.6.3"
 spec:
   selector:
     matchLabels:
       app.kubernetes.io/name: "dcgm-exporter"
-      app.kubernetes.io/version: "2.6.2"
+      app.kubernetes.io/version: "2.6.3"
   endpoints:
   - port: "metrics"
     path: "/metrics"