feat: release v1.5.1

+ support mlu type + fix go mod path + rename libcndev.so when building images + add mlu driver and mcu version info + add mlu_container metric and deprecation notice Signed-off-by: Yuting Tan <[email protected]>
Cambricon · Mar 5, 2021 · 7747617 · 7747617
1 parent 7f11950
commit 7747617
Show file tree

Hide file tree

Showing 19 changed files with 349 additions and 61 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -1,5 +1,13 @@
 # Changelog
 
+## v1.5.1
+
++ Add MLU driver, mcu and mlu type labels
++ Add mlu_container metric. Use `<metric> * on(boardid) group_right ai_mlu_container` to append k8s container info to a metric.
++ **Deprecation:** container_resource_mlu_utilization will be removed in the future
++ **Deprecation:** container_resource_mlu_memory_utilization will be removed in the future
++ **Deprecation:** container_resource_mlu_board_power will be removed in the future
+
 ## v1.5.0
 
 + Open source basic functions.
diff --git a/Makefile b/Makefile
@@ -21,8 +21,8 @@ export CC=aarch64-linux-gnu-gcc
 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/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
 
 lint:
 	golangci-lint run -v

diff --git a/README.md b/README.md
@@ -54,7 +54,7 @@ Use the following command to start the exporter.
 docker run -d \
 -p 30108:30108 \
 --privileged=true \
-cambricon-mlu-exporter:v1.5.0
+cambricon-mlu-exporter:v1.5.1
 ```
 
 Then use the following command to get the metrics.
@@ -70,7 +70,7 @@ docker run -d \
 -p 30108:30108 \
 -v examples/metrics.yaml:/var/lib/mlu-exporter/metrics.yaml \
 --privileged=true \
-cambricon-mlu-exporter:v1.5.0 \
+cambricon-mlu-exporter:v1.5.1 \
 mlu-exporter \
 --metrics-config=/var/lib/mlu-exporter/metrics.yaml \
 --metrics-path=/metrics \
@@ -124,3 +124,7 @@ kubectl apply -f examples/cambricon-mlu-exporter-sm.yaml
 Then checkout your Prometheus to get the MLU metrics.
 
 You can also set the command args described above in the MLU exporter daemonset spec. And set the metrics configuration in the MLU exporter configMap.
+
+## Upgrade Notice
+
+**Please see [changelog](CHANGELOG.md) for deprecation and breaking changes.**
diff --git a/build_image.sh b/build_image.sh
@@ -16,7 +16,7 @@
 curpath=$(dirname "$0")
 cd "$curpath" || exit 1
 
-: "${TAG:=v1.5.0}"
+: "${TAG:=v1.5.1}"
 : "${ARCH:=amd64}"
 : "${LIBCNDEV:=/usr/local/neuware/lib64/libcndev.so}"
 
@@ -65,7 +65,7 @@ if ! file "$LIBCNDEV" --dereference | grep -q "$file_arch"; then
 	exit 1
 fi
 
-cp "$LIBCNDEV" "$curpath/libs/linux/$ARCH/"
+cp "$LIBCNDEV" "$curpath/libs/linux/$ARCH/libcndev.so"
 
 echo "Building Cambricon MLU Exporter docker image."
 

diff --git a/examples/cambricon-mlu-exporter-cm.yaml b/examples/cambricon-mlu-exporter-cm.yaml
@@ -22,103 +22,164 @@ data:
         labels:
           slot: "slot"
           model: "model"
+          mluType: "mlu_type"
           sn: "boardid"
           cluster: "coregroup"
           nodeName: "nodeHostname"
+          mcu: "mcu"
+          driver: "driver"
       board_health:
         name: "board_health"
         help: "The health state of Cambricon MLU, 1 means health, 0 means sick"
         labels:
           slot: "slot"
           model: "model"
+          mluType: "mlu_type"
           sn: "boardid"
           nodeName: "nodeHostname"
+          mcu: "mcu"
+          driver: "driver"
       physical_memory_total:
         name: "mem_total_bytes"
         help: "The total physical memory of Cambricon MLU, unit is 'B'"
         labels:
           slot: "slot"
           model: "model"
+          mluType: "mlu_type"
           sn: "boardid"
           nodeName: "nodeHostname"
+          mcu: "mcu"
+          driver: "driver"
       physical_memory_used:
         name: "mem_used_bytes"
         help: "The used physical memory of Cambricon MLU, unit is 'B'"
         labels:
           slot: "slot"
           model: "model"
+          mluType: "mlu_type"
           sn: "boardid"
           nodeName: "nodeHostname"
+          mcu: "mcu"
+          driver: "driver"
       memory_utilization:
         name: "mem_utilization"
         help: "The memory utilization of Cambricon MLU, unit is '%'"
         labels:
           slot: "slot"
           model: "model"
+          mluType: "mlu_type"
           sn: "boardid"
           nodeName: "nodeHostname"
+          mcu: "mcu"
+          driver: "driver"
       board_utilization:
         name: "board_utilization"
         help: "The utilization of Cambricon MLU, unit is '%'"
         labels:
           slot: "slot"
           model: "model"
+          mluType: "mlu_type"
           sn: "boardid"
           nodeName: "nodeHostname"
+          mcu: "mcu"
+          driver: "driver"
       board_capacity:
         name: "board_capacity"
         help: "The capacity metric of Cambricon MLU, unit is 'T'"
         labels:
           slot: "slot"
           model: "model"
+          mluType: "mlu_type"
           sn: "boardid"
           nodeName: "nodeHostname"
+          mcu: "mcu"
+          driver: "driver"
       board_usage:
         name: "board_usage"
         help: "The usage metric of Cambricon MLU, unit is 'T'"
         labels:
           slot: "slot"
           model: "model"
+          mluType: "mlu_type"
           sn: "boardid"
           nodeName: "nodeHostname"
+          mcu: "mcu"
+          driver: "driver"
       core_utilization:
         name: "core"
         help: "The utilization metric of Cambricon MLU core, unit is '%'"
         labels:
           core: "core"
           slot: "slot"
           model: "model"
+          mluType: "mlu_type"
           sn: "boardid"
           nodeName: "nodeHostname"
+          mcu: "mcu"
+          driver: "driver"
       fan_speed:
         name: "fan"
         help: "Fan speed of Cambricon MLU, unit is 'rpm', '-1' means no fan on MLU"
         labels:
           slot: "slot"
           model: "model"
+          mluType: "mlu_type"
           sn: "boardid"
           nodeName: "nodeHostname"
+          mcu: "mcu"
+          driver: "driver"
       board_power:
         name: "board_power"
         help: "The power usage of Cambricon MLU, unit is 'w'"
         labels:
           slot: "slot"
           model: "model"
+          mluType: "mlu_type"
           sn: "boardid"
           nodeName: "nodeHostname"
+          mcu: "mcu"
+          driver: "driver"
+      board_version:
+        name: "board_version"
+        help: "The board version info of Cambricon MLU"
+        labels:
+          slot: "slot"
+          model: "model"
+          mluType: "mlu_type"
+          sn: "boardid"
+          nodeName: "nodeHostname"
+          mcu: "mcu"
+          driver: "driver"
     podresources:
       board_allocated:
         name: "board_allocated"
         help: "The allocated number of Cambricon MLUs in a node"
         labels:
           model: "model"
+          mluType: "mlu_type"
+          driver: "driver"
+          nodeName: "nodeHostname"
+      mlu_container:
+        name: "mlu_container"
+        help: "The k8s container info of Cambricon MLUs allocated"
+        labels:
+          slot: "mlu"
+          model: "model"
+          mluType: "mlu_type"
+          sn: "boardid"
           nodeName: "nodeHostname"
+          mcu: "mcu"
+          driver: "driver"
+          namespace: "pod_namespace"
+          pod: "pod_name"
+          container: "container_name"
       container_mlu_utilization:
         name: "container_resource_mlu_utilization"
-        help: "The utilization metric of Cambricon MLU in container, unit is '%'"
+        help: "Deprecated (this metric will be removed in the future): The utilization metric of Cambricon MLU in container, unit is '%'"
         labels:
           slot: "slot"
           model: "model"
+          mluType: "mlu_type"
           sn: "board_id"
           nodeName: "nodeHostname"
           namespace: "namespace"
@@ -130,6 +191,7 @@ data:
         labels:
           slot: "slot"
           model: "model"
+          mluType: "mlu_type"
           sn: "board_id"
           vfID: "vfid"
           nodeName: "nodeHostname"
@@ -138,10 +200,23 @@ data:
           container: "containerName"
       container_mlu_memory_utilization:
         name: "container_resource_mlu_memory_utilization"
-        help: "The memory utilization metric of Cambricon MLU in container, unit is '%'"
+        help: "Deprecated (this metric will be removed in the future): The memory utilization metric of Cambricon MLU in container, unit is '%'"
+        labels:
+          slot: "slot"
+          model: "model"
+          mluType: "mlu_type"
+          sn: "board_id"
+          nodeName: "nodeHostname"
+          namespace: "namespace"
+          pod: "pod"
+          container: "containerName"
+      container_mlu_board_power:
+        name: "container_resource_mlu_board_power"
+        help: "Deprecated (this metric will be removed in the future): The board power usage of Cambricon MLU in container, unit is 'w'"
         labels:
           slot: "slot"
           model: "model"
+          mluType: "mlu_type"
           sn: "board_id"
           nodeName: "nodeHostname"
           namespace: "namespace"

diff --git a/examples/cambricon-mlu-exporter-ds.yaml b/examples/cambricon-mlu-exporter-ds.yaml
@@ -30,7 +30,7 @@ spec:
     spec:
       containers:
       - name: cambricon-mlu-monitor
-        image: cambricon-mlu-exporter:v1.5.0
+        image: cambricon-mlu-exporter:v1.5.1
         imagePullPolicy: IfNotPresent
         command:
         - /usr/bin/mlu-exporter