github: release v2.0.11

Cambricon · Apr 26, 2024 · ecce8b2 · ecce8b2
1 parent a1430dc
commit ecce8b2
Show file tree

Hide file tree

Showing 41 changed files with 35,275 additions and 5,496 deletions.
diff --git a/.gitignore b/.gitignore
@@ -3,3 +3,5 @@ mlu-exporter
 image
 .vscode
 *.tar*
+_build/
+__pycache__/
diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
@@ -12,7 +12,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-image: 10.3.68.2:5001/cambricon/buildpack:20210507
+image: yellow.hub.cambricon.com/caip/ci/buildpack:20230712
 variables:
   GOPROXY: http://10.3.68.2:8080
 

diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -1,5 +1,74 @@
 # Changelog
 
+## v2.0.11
+
+- Remove mem-share
+
+## v2.0.10
+
+- Support sram/dram ecc err
+- Add mlu nums
+- Add heartbeat count
+
+## v2.0.9
+
+- Support dsmlu restore
+- Bump cndev to 3.9.0
+- Stop annoy metrics
+
+## v2.0.8
+
+- Support dynamic smlu monitoring
+- Support new metrics to align with dcgm
+- Support to print out version
+- Refactor to use golden test
+- Bump cndev to 3.8.0
+- Support xid errors metrics
+
+## v2.0.7
+
+- Support smlu static
+- Fix memory total and used
+
+## v2.0.6
+
+- Get rid of annoy metrics cause long latency
+- Add prometheus additional config
+
+## v2.0.5
+
+- Upgrade dependence to cndev 3.4.2
+- Eliminate annoying printing
+- Replace ioutil with os package
+- Add metric:
+  - parity_error
+
+## v2.0.4
+
+- Support mlu share mode
+- Upgrade dependence to cndev 3.4.1
+- Bump go to 1.19 and baseimage to ubuntu:20.04
+
+## v2.0.3
+
+- Add liveness/readiness probes
+- Add log level config
+- Remove beartoken in servicemonitor config
+- Report vf metrics in env share mode as in sriov mode
+
+## v2.0.1
+
+- Upgrade dependence to cndev 3.0.1
+- Deprecated and remove cnpapi dependencies
+- Refactor collect test
+- Merge vf metrics with pf metrics
+
+## v2.0.0
+
+- Upgrade dependence to cndev 3.0.0
+- Add metric:
+  - virtual_function_power_usage
+
 ## v1.6.7
 
 - Upgrade dependence to cntoolkit 2.8.2

diff --git a/Dockerfile b/Dockerfile
@@ -13,11 +13,12 @@
 # limitations under the License.
 
 ARG BUILDPLATFORM=linux/amd64
-ARG BASE_IMAGE=ubuntu:18.04
-FROM --platform=$BUILDPLATFORM golang:1.13 as build
+ARG BASE_IMAGE=ubuntu:20.04
+FROM --platform=$BUILDPLATFORM golang:1.19 as build
 ARG APT_PROXY
 ARG GOPROXY
 ARG TARGETPLATFORM
+ARG VERSION
 RUN set -ex && export http_proxy=$APT_PROXY && \
   apt-get update && \
   apt-get install -y build-essential gcc-aarch64-linux-gnu ca-certificates make

diff --git a/Makefile b/Makefile
@@ -13,6 +13,7 @@
 # limitations under the License.
 
 TARGETPLATFORM ?= linux/amd64
+VERSION ?= v1.0.0
 export GOOS := $(word 1, $(subst /, ,$(TARGETPLATFORM)))
 export GOARCH := $(word 2, $(subst /, ,$(TARGETPLATFORM)))
 export CGO_ENABLED := 1
@@ -23,14 +24,13 @@ endif
 generate:
 	mockgen -package mock -destination pkg/mock/cndev.go -mock_names=Cndev=Cndev github.com/Cambricon/mlu-exporter/pkg/cndev Cndev
 	mockgen -package mock -destination pkg/mock/podrsources.go -mock_names=PodResources=PodResources github.com/Cambricon/mlu-exporter/pkg/podresources PodResources
-	mockgen -package mock -destination pkg/mock/cnpapi.go -mock_names=Cnpapi=Cnpapi github.com/Cambricon/mlu-exporter/pkg/cnpapi Cnpapi
 	mockgen -package mock -destination pkg/mock/host.go -mock_names=Host=Host github.com/Cambricon/mlu-exporter/pkg/host Host
 
 lint:
 	golangci-lint run -v
 
 build:
-	go build  -trimpath -ldflags="-s -w" -o mlu-exporter .
+	go build -trimpath -ldflags="-s -w" -ldflags="-X 'main.version=$(VERSION)'" -o mlu-exporter .
 
 test:
 	go test -v ./...

diff --git a/README.md b/README.md
@@ -7,10 +7,11 @@ Prometheus exporter for Cambricon MLU metrics, written in Go with pluggable metr
 The prerequisites for running Cambricon MLU Exporter:
 
 - MLU270, MLU270-X5K, MLU220, MLU290, MLU370 devices
-- MLU driver >= 4.20.9
-- cntoolkit >= 2.8.2 on your building machine
+- For MLU 2xx needs driver >= 4.9.13
+- For MLU 3xx needs driver >= 4.20.9
+- For MLU 2xx、3xx needs cntoolkit >= 2.8.2 on your building machine
 
-For MLU driver version 4.9.x, please use [release v1.5.3].
+For MLU driver version before 4.9.13, please use [release v1.5.3].
 
 ## Installation and Usage
 
@@ -27,14 +28,13 @@ cd mlu-exporter
 
 Set the following environment variables if you need.
 
-| env        | description                                                                    |
-| ---------- | ------------------------------------------------------------------------------ |
-| APT_PROXY  | apt proxy address                                                              |
-| GOPROXY    | golang proxy address                                                           |
-| ARCH       | target platform architecture, amd64 or arm64, amd64 by default                 |
-| LIBCNDEV   | absolute path of the libcndev.so binary, neuware installation path by default  |
-| LIBCNPAPI  | absolute path of the libcnpapi.so binary, neuware installation path by default |
-| BASE_IMAGE | mlu exporter base image                                                        |
+| env        | description                                                                   |
+| ---------- | ----------------------------------------------------------------------------- |
+| APT_PROXY  | apt proxy address                                                             |
+| GOPROXY    | golang proxy address                                                          |
+| ARCH       | target platform architecture, amd64 or arm64, amd64 by default                |
+| LIBCNDEV   | absolute path of the libcndev.so binary, neuware installation path by default |
+| BASE_IMAGE | mlu exporter base image                                                       |
 
 Docker should be >= 17.05.0 on your building machine. If you want to cross build, make sure docker version >= 19.03.
 
@@ -67,7 +67,7 @@ docker run -d \
 --privileged=true \
 --pid=host \
 -e ENV_NODE_NAME={nodeName} \
-cambricon-mlu-exporter:v1.6.7
+cambricon-mlu-exporter:v2.0.11
 ```
 
 Then use the following command to get the metrics.
@@ -84,10 +84,11 @@ docker run -d \
 -v examples/metrics.yaml:/etc/mlu-exporter/metrics.yaml \
 --privileged=true \
 --pid=host \
-cambricon-mlu-exporter:v1.6.7 \
+cambricon-mlu-exporter:v2.0.11 \
 mlu-exporter \
 --metrics-config=/etc/mlu-exporter/metrics.yaml \
 --metrics-path=/metrics \
+--log-level=info \
 --port=30108 \
 --hostname=hostname \
 --metrics-prefix=mlu \
@@ -97,20 +98,22 @@ mlu-exporter \
 
 Command Args Description
 
-| arg            | description                                |
-| -------------- | ------------------------------------------ |
-| metrics-config | configuration file of MLU exporter metrics |
-| metrics-path   | metrics path of the exporter service       |
-| hostname       | machine hostname, or env:"ENV_NODE_NAME"   |
-| port           | exporter service port                      |
-| metrics-prefix | prefix of all metric names                 |
-| collector      | collector names, cndev by default          |
+| arg            | description                                                            |
+| -------------- | ---------------------------------------------------------------------- |
+| collector      | collector names, cndev by default                                      |
+| env-share-num  | vf numbers under env share mode, should set virtual-mode to env-share  |
+| hostname       | machine hostname, or env:"ENV_NODE_NAME"                               |
+| log-level      | set log level: trace/debug/info/warn/error/fatal/panic" default:"info" |
+| metrics-config | configuration file of MLU exporter metrics                             |
+| metrics-path   | metrics path of the exporter service                                   |
+| metrics-prefix | prefix of all metric names                                             |
+| port           | exporter service port                                                  |
+| virtual-mode   | virtual mode, default "", support dynamic-smlu, env-share              |
 
 available collectors:
 
 - cndev: collects basic MLU metrics
 - podresources: collects MLU usage metrics in containers managed by Kubernetes. For Kubernetes lower than 1.15, make sure `KubeletPodResources` [feature gate] is enabled by setting the `feature-gates` [kubelet option] in your kubelet configuration.
-- cnpapi: collects cnpapi pmu api metrics. It does not support SR-IOV. **Please note that cnpapi can only be used by one single process on a machine. Not recommended for production scenarios.**
 - host: collects host machine metrics
 
 And set the metrics configuration file passed by your metrics-config arg as you like, see examples/metrics.yaml for an example.
@@ -141,6 +144,12 @@ And if you want to create a Prometheus service monitor
 kubectl apply -f examples/cambricon-mlu-exporter-sm.yaml
 ```
 
+if you want to create a Prometheus [additional scrape configs]
+
+```shell
+kubectl create secret generic additional-scrape-configs --from-file=examples/cambricon-mlu-exporter-additional.yaml
+```
+
 Then checkout your Prometheus to get the MLU metrics.
 
 ##### Group Metrics
@@ -153,10 +162,10 @@ To attach namespace/pod/container info to another MLU metric, use `mlu_container
 mlu_power_usage * on(uuid) group_right mlu_container
 ```
 
-And for SR-IOV VFs:
+And for env-share VFs:
 
 ```text
-mlu_virtual_function_utilization * on(uuid,vf) group_right mlu_container
+mlu_utilization * on(uuid,vf) group_right mlu_container
 ```
 
 #### Metrics and Labels
@@ -174,3 +183,4 @@ For MLU370, mlu_temperature does not support cluster temperature, all cluster te
 [release v1.5.3]: https://github.com/Cambricon/mlu-exporter/releases/tag/v1.5.3
 [feature gate]: https://kubernetes.io/docs/reference/command-line-tools-reference/feature-gates/
 [kubelet option]: https://kubernetes.io/docs/reference/command-line-tools-reference/kubelet/#options
+[additional scrape configs]: https://github.com/prometheus-operator/prometheus-operator/tree/main/example/additional-scrape-configs
diff --git a/build_image.sh b/build_image.sh
@@ -16,24 +16,22 @@
 curpath=$(dirname "$0")
 cd "$curpath" || exit 1
 
-: "${TAG:=v1.6.7}"
+: "${TAG:=v2.0.11}"
 : "${ARCH:=amd64}"
 : "${LIBCNDEV:=/usr/local/neuware/lib64/libcndev.so}"
-: "${LIBCNPAPI:=/usr/local/neuware/lib64/libcnpapi.so}"
 
 case $(awk -F= '/^NAME/{print $2}' /etc/os-release) in
 "CentOS Linux")
 	BASE_IMAGE=centos:7
 	;;
 esac
 
-: "${BASE_IMAGE:=ubuntu:18.04}"
+: "${BASE_IMAGE:=ubuntu:22.04}"
 
 echo "Build environ (Can be overridden):"
 echo "TAG       = $TAG"
 echo "ARCH      = $ARCH"
 echo "LIBCNDEV  = $LIBCNDEV"
-echo "LIBCNPAPI  = $LIBCNPAPI"
 echo "APT_PROXY = $APT_PROXY"
 echo "GOPROXY   = $GOPROXY"
 echo "BASE_IMAGE   = $BASE_IMAGE"
@@ -60,12 +58,6 @@ if [[ ! -f "$LIBCNDEV" ]]; then
 	exit 1
 fi
 
-if [[ ! -f "$LIBCNPAPI" ]]; then
-	echo "Can't find libcnpapi.so at $LIBCNPAPI."
-	echo "If you want to scrape cnpapi metrics, please install Cambricon neuware, or set LIBCNPAPI environ to path of libcnpapi.so"
-	echo "Else, ignore this message."
-fi
-
 case $ARCH in
 amd64)
 	file_arch=x86-64
@@ -84,13 +76,7 @@ if ! file "$LIBCNDEV" --dereference | grep -q "$file_arch"; then
 	exit 1
 fi
 
-if [[ -f "$LIBCNPAPI" ]] && ! file "$LIBCNPAPI" --dereference | grep -q "$file_arch"; then
-	echo "$LIBCNPAPI is not for $ARCH"
-	exit 1
-fi
-
 cp "$LIBCNDEV" "$curpath/libs/linux/$ARCH/libcndev.so"
-[[ -f "$LIBCNPAPI" ]] && cp "$LIBCNPAPI" "$curpath/libs/linux/$ARCH/libcnpapi.so"
 
 echo "Building Cambricon MLU Exporter docker image."
 
@@ -100,6 +86,7 @@ echo "Building Cambricon MLU Exporter docker image."
 	--build-arg "GOPROXY=$GOPROXY" --build-arg "APT_PROXY=$APT_PROXY" \
 	--build-arg "BUILDPLATFORM=linux/$ARCH" \
 	--build-arg "BASE_IMAGE=$BASE_IMAGE" \
+	--build-arg "VERSION=$TAG" \
 	--build-arg "TARGETPLATFORM=linux/$ARCH" .
 
 [[ "$ARCH" == "$build_arch" ]] && docker save -o "image/cambricon-mlu-exporter-$ARCH.tar" \
@@ -118,4 +105,3 @@ fi
 
 echo "Image is saved at ./image/cambricon-mlu-exporter-$ARCH.tar"
 rm -f "$curpath/libs/linux/$ARCH/libcndev.so"
-rm -f "$curpath/libs/linux/$ARCH/libcnpapi.so"
diff --git a/examples/cambricon-mlu-exporter-additional.yaml b/examples/cambricon-mlu-exporter-additional.yaml
@@ -0,0 +1,17 @@
+- job_name: kube-prometheus-exporter-mlu-monitoring
+  kubernetes_sd_configs:
+  - namespaces:
+      names:
+      - kube-system
+    role: endpoints
+  metrics_path: /metrics
+  relabel_configs:
+  - source_labels: [__meta_kubernetes_endpoints_name]
+    action: keep
+    regex: "kube-prometheus-exporter-mlu-monitoring"
+  - action: replace
+    source_labels:
+    - __meta_kubernetes_endpoint_node_name
+    target_label: node
+  scheme: http
+  scrape_interval: 30s