我正在使用here中的Kafka Helm图表。 我在尝试使用水平Pod自动缩放器。
我在模板文件夹中添加了一个hpa.yaml文件，如下所示。
apiVersion: autoscaling/v2beta1
kind: HorizontalPodAutoscaler
metadata:
name: kafka-hpa
spec:
scaleTargetRef:
apiVersion: extensions/v1beta1
kind: Deployment
name: {{ include "kafka.fullname" . }}
minReplicas: {{ .Values.replicas }}
maxReplicas: 5
metrics:
- type: Resource
resource:
name: cpu
targetAverageUtilization: 50
- type: Resource
resource:
name: memory
targetAverageValue: 8000Mi
我还尝试了使用 kind：StatefulSet 的上述YAML，但是仍然存在相同的问题。
我的意图是最初拥有3个Kafka Pod，并根据上述CPU和内存targetValues将其扩展到5个。
但是，根据我的理解，hpa已部署，但是由于当前用法显示未知，因此无法读取指标。
NAME REFERENCE TARGETS MINPODS MAXPODS REPLICAS AGE
kafka-hpa Deployment/whopping-walrus-kafka <unknown>/8000Mi, <unknown>/50% 3 5 0 1h .
我是掌舵人和Kubernetes的新手，所以我认为我的理解可能存在一些问题。
我还部署了metrics-server。
$ kubectl get deployments
NAME DESIRED CURRENT UP-TO-DATE AVAILABLE AGE
metrics-server 1 1 1 1 1d
whopping-walrus-kafka-exporter 1 1 1 1 1h
豆荚输出
$ kubectl get pods
NAME READY STATUS RESTARTS AGE
metrics-server-55cbf87bbb-vm2v5 1/1 Running 0 15m
whopping-walrus-kafka-0 1/1 Running 1 1h
whopping-walrus-kafka-1 1/1 Running 0 1h
whopping-walrus-kafka-2 1/1 Running 0 1h
whopping-walrus-kafka-exporter-5c66b5b4f9-mv5kv 1/1 Running 1 1h
whopping-walrus-zookeeper-0 1/1 Running 0 1h
我希望 whopping-walrus-kafka 窗格在加载时最多可扩展到5个，但是没有对应的部署。
StatefulSet输出
$ kubectl get statefulset
NAME DESIRED CURRENT AGE
original-bobcat-kafka 3 2 2m
original-bobcat-zookeeper 1 1 2m
hpa.yaml中的kind为 StatefulSet 时，描述hpa的输出。
$ kubectl describe hpa
Name: kafka-hpa
Namespace: default
Labels: <none>
Annotations: <none>
CreationTimestamp: Fri, 18 Jan 2019 12:13:59 +0530
Reference: StatefulSet/original-bobcat-kafka
Metrics: ( current / target )
resource memory on pods: <unknown> / 8000Mi
resource cpu on pods (as a percentage of request): <unknown> / 5%
Min replicas: 3
Max replicas: 5
Conditions:
Type Status Reason Message
---- ------ ------ -------
AbleToScale False FailedGetScale the HPA controller was unable to get the target's current scale: no matches for kind "StatefulSet" in group "extensions"
Events:
Type Reason Age From Message
---- ------ ---- ---- -------
Warning FailedGetScale 15s (x17 over 8m) horizontal-pod-autoscaler no matches for kind "StatefulSet" in group "extensions"
hpa.yaml中的kind为部署时，描述hpa的输出。
$ kubectl describe hpa
Name: kafka-hpa
Namespace: default
Labels: <none>
Annotations: <none>
CreationTimestamp: Fri, 18 Jan 2019 12:30:07 +0530
Reference: Deployment/good-elephant-kafka
Metrics: ( current / target )
resource memory on pods: <unknown> / 8000Mi
resource cpu on pods (as a percentage of request): <unknown> / 5%
Min replicas: 3
Max replicas: 5
Conditions:
Type Status Reason Message
---- ------ ------ -------
AbleToScale False FailedGetScale the HPA controller was unable to get the target's current scale: could not fetch the scale for deployments.extensions good-elephant-kafka: deployments/scale.extensions "good-elephant-kafka" not found
Events:
Type Reason Age From Message
---- ------ ---- ---- -------
Warning FailedGetScale 9s horizontal-pod-autoscaler could not fetch the scale for deployments.extensions good-elephant-kafka: deployments/scale.extensions "good-elephant-kafka" not found
指标服务器容器的输出
$ kubectl describe pods metrics-server-55cbf87bbb-vm2v5
Name: metrics-server-55cbf87bbb-vm2v5
Namespace: default
Node: docker-for-desktop/192.168.65.3
Start Time: Fri, 18 Jan 2019 11:26:33 +0530
Labels: app=metrics-server
pod-template-hash=1176943666
release=metrics-server
Annotations: <none>
Status: Running
IP: 10.1.0.119
Controlled By: ReplicaSet/metrics-server-55cbf87bbb
Containers:
metrics-server:
Container ID: docker://ee4b3d9ed1b15c2c8783345b0ffbbc565ad25f1493dec0148f245c9581443631
Image: gcr.io/google_containers/metrics-server-amd64:v0.3.1
Image ID: docker-pullable://gcr.io/google_containers/metrics-server-amd64@sha256:78938f933822856f443e6827fe5b37d6cc2f74ae888ac8b33d06fdbe5f8c658b
Port: <none>
Host Port: <none>
Command:
/metrics-server
--kubelet-insecure-tls
--kubelet-preferred-address-types=InternalIP
--logtostderr
State: Running
Started: Fri, 18 Jan 2019 11:26:35 +0530
Ready: True
Restart Count: 0
Environment: <none>
Mounts:
/var/run/secrets/kubernetes.io/serviceaccount from metrics-server-token-d2g7b (ro)
Conditions:
Type Status
Initialized True
Ready True
PodScheduled True
Volumes:
metrics-server-token-d2g7b:
Type: Secret (a volume populated by a Secret)
SecretName: metrics-server-token-d2g7b
Optional: false
QoS Class: BestEffort
Node-Selectors: <none>
Tolerations: node.kubernetes.io/not-ready:NoExecute for 300s
node.kubernetes.io/unreachable:NoExecute for 300s
Events: <none>
如果我在某个地方出错，请大家也清除我的理解。
在此方面的帮助将不胜感激。
答案 0 :(得分：0)
您需要在metrics-server部署文件中添加以下命令：
containers:
- command:
- /metrics-server
- --metric-resolution=30s
- --kubelet-insecure-tls
- --kubelet-preferred-address-types=InternalIP
name: metrics-server
我认为metrics-server无法找到带有InternalIP的kubelet，因此出现了问题。有关更多信息，请检查我下面的答案以逐步设置HPA。
答案 1 :(得分：0)
我执行了一些操作，类似于上面@PrafullLadha提到的操作。
修改了指标服务器部署文件并添加了以下代码：
containers:
- command:
- /metrics-server
- --metric-resolution=30s
- --kubelet-insecure-tls
- --kubelet-preferred-address-types=InternalIP`
此外，取消注释statefulset.yaml文件中的以下部分
resources:
requests:
cpu: 200m
memory: 256Mi
在上面效果很好。
答案 2 :(得分：0)
如果您的部署到目前为止没有节点，并且群集中可用的节点不满足您的资源要求，那么也会发生此错误。在这种情况下，显然没有可用的指标。