Pod autoscaling

Makefile

@@ -117,6 +117,7 @@ test-examples:
 ###############
 
 ci-build-images:
+	@./build/build-image.sh images/manager manager
 	@./build/build-image.sh images/spark-base spark-base
 	@./build/build-image.sh images/tf-base tf-base
 	@./build/build-image.sh images/tf-base-gpu tf-base-gpu
@@ -135,9 +136,10 @@ ci-build-images:
 	@./build/build-image.sh images/argo-executor argo-executor
 	@./build/build-image.sh images/python-packager python-packager
 	@./build/build-image.sh images/cluster-autoscaler cluster-autoscaler
-	@./build/build-image.sh images/manager manager
+	@./build/build-image.sh images/metrics-server metrics-server
 
 ci-push-images:
+	@./build/push-image.sh manager
 	@./build/push-image.sh spark
 	@./build/push-image.sh spark-operator
 	@./build/push-image.sh tf-train
@@ -153,7 +155,8 @@ ci-push-images:
 	@./build/push-image.sh argo-executor
 	@./build/push-image.sh python-packager
 	@./build/push-image.sh cluster-autoscaler
-	@./build/push-image.sh manager
+	@./build/push-image.sh metrics-server
+
 
 ci-build-cli:
 	@./build/cli.sh

README.md

@@ -26,8 +26,8 @@ Cortex is actively maintained by Cortex Labs. We're a venture-backed team of inf
   model: s3://my-bucket/my-model.zip
   request_handler: handler.py
   compute:
-    replicas: 4
-    gpu: 2
+    min_replicas: 5
+    max_replicas: 20
 ```
 
 **Customize** request handling (optional):

cli/cmd/get.go

@@ -443,16 +443,16 @@ func describeAPI(name string, resourcesRes *schema.GetResourcesResponse) (string
 	}
 
 	out := titleStr("Summary")
-	out += "Status:            " + groupStatus.Message() + "\n"
+	out += "Status:               " + groupStatus.Message() + "\n"
 	if ctxAPIStatus != nil {
-		out += fmt.Sprintf("Updated replicas:  %d/%d ready\n", ctxAPIStatus.ReadyUpdated, ctxAPIStatus.RequestedReplicas)
+		out += fmt.Sprintf("Up-to-date replicas:  %d ready\n", ctxAPIStatus.ReadyUpdated)
 	}
 	if staleReplicas != 0 {
-		out += fmt.Sprintf("Stale replicas:    %d ready\n", staleReplicas)
+		out += fmt.Sprintf("Stale replicas:       %d ready\n", staleReplicas)
 	}
-	out += "Created at:        " + libtime.LocalTimestamp(groupStatus.Start) + "\n"
+	out += "Created at:           " + libtime.LocalTimestamp(groupStatus.Start) + "\n"
 	if groupStatus.ActiveStatus != nil && groupStatus.ActiveStatus.Start != nil {
-		out += "Refreshed at:      " + libtime.LocalTimestamp(groupStatus.ActiveStatus.Start) + "\n"
+		out += "Refreshed at:         " + libtime.LocalTimestamp(groupStatus.ActiveStatus.Start) + "\n"
 	}
 
 	out += titleStr("Endpoint")

cortex.sh

@@ -131,6 +131,7 @@ export CORTEX_IMAGE_PYTHON_PACKAGER="${CORTEX_IMAGE_PYTHON_PACKAGER:-cortexlabs/
 export CORTEX_IMAGE_TF_SERVE_GPU="${CORTEX_IMAGE_TF_SERVE_GPU:-cortexlabs/tf-serve-gpu:$CORTEX_VERSION_STABLE}"
 export CORTEX_IMAGE_TF_TRAIN_GPU="${CORTEX_IMAGE_TF_TRAIN_GPU:-cortexlabs/tf-train-gpu:$CORTEX_VERSION_STABLE}"
 export CORTEX_IMAGE_CLUSTER_AUTOSCALER="${CORTEX_IMAGE_CLUSTER_AUTOSCALER:-cortexlabs/cluster-autoscaler:$CORTEX_VERSION_STABLE}"
+export CORTEX_IMAGE_METRICS_SERVER="${CORTEX_IMAGE_METRICS_SERVER:-cortexlabs/metrics-server:$CORTEX_VERSION_STABLE}"
 
 export CORTEX_ENABLE_TELEMETRY="${CORTEX_ENABLE_TELEMETRY:-""}"
 
@@ -181,6 +182,7 @@ function install_cortex() {
     -e CORTEX_IMAGE_TF_SERVE_GPU=$CORTEX_IMAGE_TF_SERVE_GPU \
     -e CORTEX_IMAGE_TF_TRAIN_GPU=$CORTEX_IMAGE_TF_TRAIN_GPU \
     -e CORTEX_IMAGE_CLUSTER_AUTOSCALER=$CORTEX_IMAGE_CLUSTER_AUTOSCALER \
+    -e CORTEX_IMAGE_METRICS_SERVER=$CORTEX_IMAGE_METRICS_SERVER \
     -e CORTEX_ENABLE_TELEMETRY=$CORTEX_ENABLE_TELEMETRY \
     $CORTEX_IMAGE_MANAGER
 }

dev/load.sh

@@ -0,0 +1,42 @@
+#!/bin/bash
+
+# Copyright 2019 Cortex Labs, Inc.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+
+SLEEP="0.01"
+
+URL="https://ac47cdf2e99cb11e9ade10693263c378-1155928797.us-west-2.elb.amazonaws.com/iris/iris-type"
+DATA='{ "samples": [ { "sepal_length": 5.2, "sepal_width": 3.6, "petal_length": 1.4, "petal_width": 0.3 } ] }'
+
+# URL="https://ac47cdf2e99cb11e9ade10693263c378-1155928797.us-west-2.elb.amazonaws.com/insurance/cost"
+# DATA='{ "samples": [ { "age": 22, "bmi": 25, "children": 0, "region": "northeast", "sex": "female", "smoker": "no" } ] }'
+
+trap ctrl_c INT
+function ctrl_c() {
+  echo ""
+  exit 0
+}
+
+function make_request() {
+  curl --silent --show-error -k -X POST -H "Content-Type: application/json" -d "${DATA}" "${URL}"
+}
+
+resp=$(make_request)
+echo -n "."
+
+while eval "sleep ${SLEEP}"; do
+  resp=$(make_request)
+  echo -n "."
+done

dev/registry.sh

@@ -35,6 +35,7 @@ function ecr_login() {
 }
 
 function create_registry() {
+  aws ecr create-repository --repository-name=cortexlabs/manager --region=$REGISTRY_REGION || true
   aws ecr create-repository --repository-name=cortexlabs/argo-controller --region=$REGISTRY_REGION || true
   aws ecr create-repository --repository-name=cortexlabs/argo-executor --region=$REGISTRY_REGION || true
   aws ecr create-repository --repository-name=cortexlabs/fluentd --region=$REGISTRY_REGION || true
@@ -50,7 +51,7 @@ function create_registry() {
   aws ecr create-repository --repository-name=cortexlabs/tf-train-gpu --region=$REGISTRY_REGION || true
   aws ecr create-repository --repository-name=cortexlabs/tf-serve-gpu --region=$REGISTRY_REGION || true
   aws ecr create-repository --repository-name=cortexlabs/cluster-autoscaler --region=$REGISTRY_REGION || true
-  aws ecr create-repository --repository-name=cortexlabs/manager --region=$REGISTRY_REGION || true
+  aws ecr create-repository --repository-name=cortexlabs/metrics-server --region=$REGISTRY_REGION || true
 }
 
 ### HELPERS ###
@@ -116,6 +117,8 @@ if [ "$cmd" = "create" ]; then
 
 elif [ "$cmd" = "update" ]; then
   if [ "$env" != "dev" ]; then
+    build_and_push $ROOT/images/manager manager latest
+
     cache_builder $ROOT/images/spark-base spark-base
     build_base $ROOT/images/spark-base spark-base
     build_base $ROOT/images/tf-base tf-base
@@ -136,7 +139,7 @@ elif [ "$cmd" = "update" ]; then
     build_and_push $ROOT/images/tf-serve-gpu tf-serve-gpu latest
     build_and_push $ROOT/images/python-packager python-packager latest
     build_and_push $ROOT/images/cluster-autoscaler cluster-autoscaler latest
-    build_and_push $ROOT/images/manager manager latest
+    build_and_push $ROOT/images/metrics-server metrics-server latest
   fi
 
   build_and_push $ROOT/images/spark spark latest

docs/apis/apis.md

@@ -9,9 +9,12 @@ Serve models at scale and use them to build smarter applications.
   name: <string>  # API name (required)
   model: <string>  # path to a zipped model dir (e.g. s3://my-bucket/model.zip)
   compute:
-    replicas: <int>  # number of replicas to launch (default: 1)
-    cpu: <string>  # CPU request per replica (default: Null)
-    gpu: <string>  # gpu request per replica (default: Null)
+    min_replicas: <int>  # minimum number of replicas (default: 1)
+    max_replicas: <int>  # maximum number of replicas (default: 100)
+    init_replicas: <int>  # initial number of replicas (default: <min_replicas>)
+    target_cpu_utilization: <int>  # CPU utilization threshold (as a percentage) to trigger scaling (default: 80)
+    cpu: <string>  # CPU request per replica (default: 200m)
+    gpu: <string>  # gpu request per replica (default: 0)
     mem: <string>  # memory request per replica (default: Null)
 ```
 
@@ -24,8 +27,9 @@ See [packaging models](packaging-models.md) for how to create the zipped model.
   name: my-api
   model: s3://my-bucket/my-model.zip
   compute:
-    replicas: 3
-    gpu: 2
+    min_replicas: 5
+    max_replicas: 20
+    cpu: "1"
 ```
 
 ## Integration

docs/apis/tutorial.md

@@ -33,8 +33,6 @@ Add to `cortex.yaml`:
 - kind: api
   name: iris-type
   model: s3://cortex-examples/iris-tensorflow.zip
-  compute:
-    replicas: 3
 ```
 
 ### Deploy the API

docs/cluster/config.md

@@ -52,6 +52,7 @@ export CORTEX_IMAGE_TF_TRAIN_GPU="cortexlabs/tf-train-gpu:master"
 export CORTEX_IMAGE_TF_SERVE_GPU="cortexlabs/tf-serve-gpu:master"
 export CORTEX_IMAGE_PYTHON_PACKAGER="cortexlabs/python-packager:master"
 export CORTEX_IMAGE_CLUSTER_AUTOSCALER="cortexlabs/cluster-autoscaler:master"
+export CORTEX_IMAGE_METRICS_SERVER="cortexlabs/metrics-server:master"
 
 # Flag to enable collecting error reports and usage stats. If flag is not set to either "true" or "false", you will be prompted.
 export CORTEX_ENABLE_TELEMETRY=""

docs/cluster/development.md

@@ -86,6 +86,7 @@ export CORTEX_IMAGE_TF_TRAIN_GPU="XXXXXXXX.dkr.ecr.us-west-2.amazonaws.com/corte
 export CORTEX_IMAGE_TF_TRANSFORM="XXXXXXXX.dkr.ecr.us-west-2.amazonaws.com/cortexlabs/tf-transform:latest"
 export CORTEX_IMAGE_PYTHON_PACKAGER="XXXXXXXX.dkr.ecr.us-west-2.amazonaws.com/cortexlabs/python-packager:latest"
 export CORTEX_IMAGE_CLUSTER_AUTOSCALER="XXXXXXXX.dkr.ecr.us-west-2.amazonaws.com/cortexlabs/cluster-autoscaler:latest"
+export CORTEX_IMAGE_METRICS_SERVER="XXXXXXXX.dkr.ecr.us-west-2.amazonaws.com/cortexlabs/metrics-server:latest"
 
 export AWS_ACCESS_KEY_ID="XXXXXX"
 export AWS_SECRET_ACCESS_KEY="XXXXXX"

docs/pipelines/apis.md

@@ -9,9 +9,12 @@ Serve models at scale and use them to build smarter applications.
   name: <string>  # API name (required)
   model: <string>  # reference to a model (e.g. @dnn) or path to a zipped model dir (e.g. s3://my-bucket/model.zip)
   compute:
-    replicas: <int>  # number of replicas to launch (default: 1)
-    cpu: <string>  # CPU request per replica (default: Null)
-    gpu: <string>  # gpu request per replica (default: Null)
+    min_replicas: <int>  # minimum number of replicas (default: 1)
+    max_replicas: <int>  # maximum number of replicas (default: 100)
+    init_replicas: <int>  # initial number of replicas (default: <min_replicas>)
+    target_cpu_utilization: <int>  # CPU utilization threshold (as a percentage) to trigger scaling (default: 80)
+    cpu: <string>  # CPU request per replica (default: 200m)
+    gpu: <string>  # gpu request per replica (default: 0)
     mem: <string>  # memory request per replica (default: Null)
 ```
 
@@ -24,8 +27,9 @@ See [packaging models](packaging-models.md) for how to create a zipped model (on
   name: my-api
   model: @dnn
   compute:
-    replicas: 3
-    gpu: 2
+    min_replicas: 5
+    max_replicas: 20
+    cpu: "1"
 ```
 
 ## Integration

docs/pipelines/models.md

@@ -41,9 +41,9 @@ Train TensorFlow models at scale.
     throttle_secs: <int>  # do not re-evaluate unless the last evaluation was started at least this many seconds ago (default: 600)
 
   compute:         # Resources for training and evaluations steps (TensorFlow)
-    cpu: <string>  # CPU request (default: Null)
+    cpu: <string>  # CPU request (default: 200m)
     mem: <string>  # memory request (default: Null)
-    gpu: <string>  # GPU request (default: Null)
+    gpu: <string>  # GPU request (default: 0)
 
   dataset_compute:    # Resources for constructing training dataset (Spark)
     executors: <int>  # number of spark executors (default: 1)

docs/pipelines/tutorial.md

@@ -86,8 +86,6 @@ Add to `cortex.yaml`:
 - kind: api
   name: iris-type
   model: @dnn
-  compute:
-    replicas: 2
 ```
 
 ## Deploy the application

examples/iris/cortex.yaml

@@ -4,5 +4,3 @@
 - kind: api
   name: iris-type
   model: s3://cortex-examples/iris-tensorflow.zip
-  compute:
-    replicas: 3

examples/pipelines/insurance/resources/apis.yaml

@@ -1,5 +1,3 @@
 - kind: api
   name: cost
   model: @dnn
-  compute:
-    replicas: 1

examples/pipelines/iris/cortex.yaml

@@ -24,5 +24,3 @@
 - kind: api
   name: iris-type
   model: @dnn
-  compute:
-    replicas: 1

examples/pipelines/mnist/resources/apis.yaml

@@ -1,17 +1,11 @@
 - kind: api
   name: dnn-classifier
   model: @dnn
-  compute:
-    replicas: 1
 
 - kind: api
   name: conv-classifier
   model: @conv
-  compute:
-    replicas: 1
 
 - kind: api
   name: t2t-classifier
   model: @t2t
-  compute:
-    replicas: 1

examples/pipelines/movie-ratings/resources/apis.yaml

@@ -1,5 +1,3 @@
 - kind: api
   name: ratings
   model: @basic_embedding
-  compute:
-    replicas: 1

examples/pipelines/reviews/resources/apis.yaml

@@ -1,17 +1,11 @@
 - kind: api
   name: sentiment-dnn
   model: @sentiment_dnn
-  compute:
-    replicas: 1
 
 - kind: api
   name: sentiment-linear
   model: @sentiment_linear
-  compute:
-    replicas: 1
 
 - kind: api
   name: sentiment-t2t
   model: @transformer
-  compute:
-    replicas: 1

examples/pipelines/wine/resources/apis.yaml

@@ -1,5 +1,3 @@
 - kind: api
   name: quality
   model: @dnn
-  compute:
-    replicas: 1

go.mod

@@ -20,16 +20,17 @@ go 1.12
 require (
 	github.com/GoogleCloudPlatform/spark-on-k8s-operator v0.0.0-20181208011959-62db1d66dafa
 	github.com/argoproj/argo v2.3.0+incompatible
-	github.com/aws/aws-sdk-go v1.20.8
+	github.com/aws/aws-sdk-go v1.20.12
 	github.com/cortexlabs/yaml v0.0.0-20190626164117-202ab3a3d475
 	github.com/davecgh/go-spew v1.1.1
 	github.com/emicklei/go-restful v2.9.6+incompatible // indirect
+	github.com/ghodss/yaml v1.0.0
 	github.com/go-openapi/spec v0.19.2 // indirect
 	github.com/gogo/protobuf v1.2.1 // indirect
 	github.com/google/btree v1.0.0 // indirect
 	github.com/google/gofuzz v1.0.0 // indirect
 	github.com/googleapis/gnostic v0.3.0 // indirect
-	github.com/gorilla/mux v1.7.2
+	github.com/gorilla/mux v1.7.3
 	github.com/gorilla/websocket v1.4.0
 	github.com/gregjones/httpcache v0.0.0-20190611155906-901d90724c79 // indirect
 	github.com/imdario/mergo v0.3.7 // indirect

go.sum

@@ -13,8 +13,8 @@ github.com/PuerkitoBio/urlesc v0.0.0-20170810143723-de5bf2ad4578/go.mod h1:uGdko
 github.com/argoproj/argo v2.3.0+incompatible h1:L1OYZ86Q7NK19ahdl/eJOq78Mlf52wUKGmp7VDNQVz8=
 github.com/argoproj/argo v2.3.0+incompatible/go.mod h1:KJ0MB+tuhtAklR4jkPM10mIZXfRA0peTYJ1sLUnFLVU=
 github.com/armon/consul-api v0.0.0-20180202201655-eb2c6b5be1b6/go.mod h1:grANhF5doyWs3UAsr3K4I6qtAmlQcZDesFNEHPZAzj8=
-github.com/aws/aws-sdk-go v1.20.8 h1:oXnNFH/8PosUvHXqZHYLUn6LCSzHZ/34H52vrB4YhpA=
-github.com/aws/aws-sdk-go v1.20.8/go.mod h1:KmX6BPdI08NWTb3/sm4ZGu5ShLoqVDhKgpiN924inxo=
+github.com/aws/aws-sdk-go v1.20.12 h1:xV7xfLSkiqd7JOnLlfER+Jz8kI98rAGJvtXssYkCRs4=
+github.com/aws/aws-sdk-go v1.20.12/go.mod h1:KmX6BPdI08NWTb3/sm4ZGu5ShLoqVDhKgpiN924inxo=
 github.com/coreos/etcd v3.3.10+incompatible/go.mod h1:uF7uidLiAD3TWHmW31ZFd/JWoc32PjwdhPthX9715RE=
 github.com/coreos/go-etcd v2.0.0+incompatible/go.mod h1:Jez6KQU2B/sWsbdaef3ED8NzMklzPG4d5KIOhIy30Tk=
 github.com/coreos/go-semver v0.2.0/go.mod h1:nnelYz7RCh+5ahJtPPxZlU+153eP4D4r3EedlOD2RNk=
@@ -30,6 +30,8 @@ github.com/emicklei/go-restful v2.9.6+incompatible h1:tfrHha8zJ01ywiOEC1miGY8st1
 github.com/emicklei/go-restful v2.9.6+incompatible/go.mod h1:otzb+WCGbkyDHkqmQmT5YD2WR4BBwUdeQoFo8l/7tVs=
 github.com/fsnotify/fsnotify v1.4.7/go.mod h1:jwhsz4b93w/PPRr/qN1Yymfu8t87LnFCMoQvtojpjFo=
 github.com/ghodss/yaml v0.0.0-20150909031657-73d445a93680/go.mod h1:4dBDuWmgqj2HViK6kFavaiC9ZROes6MMH2rRYeMEF04=
+github.com/ghodss/yaml v1.0.0 h1:wQHKEahhL6wmXdzwWG11gIVCkOv05bNOh+Rxn0yngAk=
+github.com/ghodss/yaml v1.0.0/go.mod h1:4dBDuWmgqj2HViK6kFavaiC9ZROes6MMH2rRYeMEF04=
 github.com/go-openapi/jsonpointer v0.0.0-20160704185906-46af16f9f7b1/go.mod h1:+35s3my2LFTysnkMfxsJBAMHj/DoqoB9knIWoYG/Vk0=
 github.com/go-openapi/jsonpointer v0.19.2 h1:A9+F4Dc/MCNB5jibxf6rRvOvR/iFgQdyNx9eIhnGqq0=
 github.com/go-openapi/jsonpointer v0.19.2/go.mod h1:3akKfEdA7DF1sugOqz1dVQHBcuDBPKZGEoHC/NkiQRg=
@@ -55,8 +57,8 @@ github.com/google/gofuzz v1.0.0/go.mod h1:dBl0BpW6vV/+mYPU4Po3pmUjxk6FQPldtuIdl/
 github.com/googleapis/gnostic v0.0.0-20170426233943-68f4ded48ba9/go.mod h1:sJBsCZ4ayReDTBIg8b9dl28c5xFWyhBTVRp3pOg5EKY=
 github.com/googleapis/gnostic v0.3.0 h1:CcQijm0XKekKjP/YCz28LXVSpgguuB+nCxaSjCe09y0=
 github.com/googleapis/gnostic v0.3.0/go.mod h1:sJBsCZ4ayReDTBIg8b9dl28c5xFWyhBTVRp3pOg5EKY=
-github.com/gorilla/mux v1.7.2 h1:zoNxOV7WjqXptQOVngLmcSQgXmgk4NMz1HibBchjl/I=
-github.com/gorilla/mux v1.7.2/go.mod h1:1lud6UwP+6orDFRuTfBEV8e9/aOM/c4fVVCaMa2zaAs=
+github.com/gorilla/mux v1.7.3 h1:gnP5JzjVOuiZD07fKKToCAOjS0yOpj/qPETTXCCS6hw=
+github.com/gorilla/mux v1.7.3/go.mod h1:1lud6UwP+6orDFRuTfBEV8e9/aOM/c4fVVCaMa2zaAs=
 github.com/gorilla/websocket v1.4.0 h1:WDFjx/TMzVgy9VdMMQi2K2Emtwi2QcUQsztZ/zLaH/Q=
 github.com/gorilla/websocket v1.4.0/go.mod h1:E7qHFY5m1UJ88s3WnNqhKjPHQ0heANvMoAMk2YaljkQ=
 github.com/gregjones/httpcache v0.0.0-20190611155906-901d90724c79 h1:+ngKgrYPPJrOjhax5N+uePQ0Fh1Z7PheYoUI/0nzkPA=
@@ -164,7 +166,6 @@ golang.org/x/tools v0.0.0-20181011042414-1f849cf54d09/go.mod h1:n7NCudcB/nEzxVGm
 golang.org/x/tools v0.0.0-20190614205625-5aca471b1d59/go.mod h1:/rFqwRUd4F7ZHNgwSSTFct+R/Kf4OFW1sUzUTQQTgfc=
 google.golang.org/appengine v1.4.0 h1:/wp5JvzpHIxhs/dumFmF7BXTf3Z+dd4uXta4kVyO508=
 google.golang.org/appengine v1.4.0/go.mod h1:xpcJRLb0r/rnEns0DIKYYv+WjYCduHsrkT7/EB5XEv4=
-gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405 h1:yhCVgyC4o1eVCa2tZl7eS0r+SDo693bJlVdllGtEeKM=
 gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0=
 gopkg.in/check.v1 v1.0.0-20180628173108-788fd7840127 h1:qIbj1fsPNlZgppZ+VLlY7N33q108Sa+fhmuc+sWQYwY=
 gopkg.in/check.v1 v1.0.0-20180628173108-788fd7840127/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0=

images/metrics-server/Dockerfile

@@ -0,0 +1 @@
+FROM k8s.gcr.io/metrics-server-amd64:v0.3.2

manager/install_cortex.sh

@@ -168,5 +168,6 @@ envsubst < manifests/nginx.yaml | kubectl apply -f - >/dev/null
 envsubst < manifests/fluentd.yaml | kubectl apply -f - >/dev/null
 envsubst < manifests/operator.yaml | kubectl apply -f - >/dev/null
 envsubst < manifests/cluster-autoscaler.yaml | kubectl apply -f - >/dev/null
+envsubst < manifests/metrics-server.yaml | kubectl apply -f - >/dev/null
 
 validate_cortex