Skip to content

Commit 8f228d1

Browse files
windsonseawindsonsea
committed
[zh] sync blog: hpa-container-resource-metric.md
1 parent d659949 commit 8f228d1

File tree

1 file changed

+209
-0
lines changed

1 file changed

+209
-0
lines changed

‎content/zh-cn/blog/_posts/2023-05-02-hpa-container-resource-metric.md

Lines changed: 209 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,209 @@
1+
---
2+
layout: blog
3+
title: "Kubernetes 1.27：HorizontalPodAutoscaler ContainerResource 类型指标进阶至 Beta"
4+
date: 2023-05-02T12:00:00+0800
5+
slug: hpa-container-resource-metric
6+
---
7+
<!--
8+
layout: blog
9+
title: "Kubernetes 1.27: HorizontalPodAutoscaler ContainerResource type metric moves to beta"
10+
date: 2023-05-02T12:00:00+0800
11+
slug: hpa-container-resource-metric
12+
-->
13+
14+
<!--
15+
**Author:** [Kensei Nakada](https://github.com/sanposhiho) (Mercari)
16+
-->
17+
**作者:** [Kensei Nakada](https://github.com/sanposhiho) (Mercari)
18+
19+
**译者:** [Michael Yao](https://github.com/windsonsea) (DaoCloud)
20+
21+
<!--
22+
Kubernetes 1.20 introduced the [`ContainerResource` type metric](/docs/tasks/run-application/horizontal-pod-autoscale/#container-resource-metrics)
23+
in HorizontalPodAutoscaler (HPA).
24+
25+
In Kubernetes 1.27, this feature moves to beta and the corresponding feature gate (`HPAContainerMetrics`) gets enabled by default.
26+
-->
27+
Kubernetes 1.20 在 HorizontalPodAutoscaler (HPA) 中引入了
28+
[`ContainerResource` 类型指标](/zh-cn/docs/tasks/run-application/horizontal-pod-autoscale/#container-resource-metrics)。
29+
30+
在 Kubernetes 1.27 中，此特性进阶至 Beta，相应的特性门控 (`HPAContainerMetrics`) 默认被启用。
31+
32+
<!--
33+
## What is the ContainerResource type metric
34+
35+
The ContainerResource type metric allows us to configure the autoscaling based on resource usage of individual containers.
36+
37+
In the following example, the HPA controller scales the target
38+
so that the average utilization of the cpu in the application container of all the pods is around 60%.
39+
(See [the algorithm details](/docs/tasks/run-application/horizontal-pod-autoscale/#algorithm-details)
40+
to know how the desired replica number is calculated exactly)
41+
-->
42+
## 什么是 ContainerResource 类型指标
43+
44+
ContainerResource 类型指标允许我们根据各个容器的资源使用量来配置自动扩缩。
45+
46+
在下面的示例中，HPA 控制器扩缩目标，以便所有 Pod 的应用程序容器的 CPU 平均利用率约为 60％
47+
（请参见[算法详情](/zh-cn/docs/tasks/run-application/horizontal-pod-autoscale/#algorithm-details)以了解预期副本数的确切计算方式）。
48+
49+
```yaml
50+
type: ContainerResource
51+
containerResource:
52+
name: cpu
53+
container: application
54+
target:
55+
type: Utilization
56+
averageUtilization: 60
57+
```
58+
59+
<!--
60+
## The difference from the Resource type metric
61+
62+
HPA already had a [Resource type metric](/docs/tasks/run-application/horizontal-pod-autoscale/#support-for-resource-metrics).
63+
64+
You can define the target resource utilization like the following,
65+
and then HPA will scale up/down the replicas based on the current utilization.
66+
-->
67+
## 与 Resource 类型指标的区别
68+
69+
HPA 已具有 [Resource 类型指标](/zh-cn/docs/tasks/run-application/horizontal-pod-autoscale/#support-for-resource-metrics)。
70+
71+
你可以定义如下的目标资源利用率，然后 HPA 将基于当前利用率扩缩副本。
72+
73+
```yaml
74+
type: Resource
75+
resource:
76+
name: cpu
77+
target:
78+
type: Utilization
79+
averageUtilization: 60
80+
```
81+
82+
<!--
83+
But, this Resource type metric refers to the average utilization of the **Pods**.
84+
85+
In case a Pod has multiple containers, the utilization calculation would be:
86+
-->
87+
但这个 Resource 类型指标指的是 **Pod** 的平均利用率。
88+
89+
如果一个 Pod 有多个容器，则利用率计算公式为：
90+
91+
<!--
92+
```
93+
sum{the resource usage of each container} / sum{the resource request of each container}
94+
```
95+
-->
96+
```
97+
sum{每个容器的资源使用量} / sum{每个容器的资源请求}
98+
```
99+
100+
<!--
101+
The resource utilization of each container may not have a direct correlation or may grow at different rates as the load changes.
102+
-->
103+
每个容器的资源利用率可能没有直接关系，或可能随着负载变化而以不同的速度增长。
104+
105+
<!--
106+
For example:
107+
- A sidecar container is only providing an auxiliary service such as log shipping.
108+
If the application does not log very frequently or does not produce logs in its hotpath
109+
then the usage of the log shipper will not grow.
110+
- A sidecar container which provides authentication. Due to heavy caching
111+
the usage will only increase slightly when the load on the main container increases.
112+
In the current blended usage calculation approach this usually results in
113+
the HPA not scaling up the deployment because the blended usage is still low.
114+
- A sidecar may be injected without resources set which prevents scaling
115+
based on utilization. In the current logic the HPA controller can only scale
116+
on absolute resource usage of the pod when the resource requests are not set.
117+
-->
118+
例如：
119+
120+
- 边车容器仅提供日志传输这类辅助服务。
121+
如果应用程序不经常记录日志或在其频繁执行的路径中不生成日志，则日志发送器的使用量不会增长。
122+
- 提供身份验证的边车容器。由于重度缓存，当主要容器的负载增加时，使用量只会略微增加。
123+
在当前的混合用量计算方法中，这通常导致 HPA 不会对 Deployment 向上扩容，因为混合的使用量仍然很低。
124+
- 边车可能在未设置资源的情况下被注入，这会阻止基于利用率进行扩缩。
125+
在当前的逻辑中，当未设置资源请求时，HPA 控制器只能根据 Pod 的绝对资源使用量进行扩缩。
126+
127+
<!--
128+
And, in such case, if only one container's resource utilization goes high,
129+
the Resource type metric may not suggest scaling up.
130+
131+
So, for the accurate autoscaling, you may want to use the ContainerResource type metric for such Pods instead.
132+
-->
133+
在这种情况下，如果仅有一个容器的资源利用率增加，则 Resource 类型指标可能不会建议扩容。
134+
135+
因此，为了实现准确的自动扩缩，你可能需要改为使用 ContainerResource 类型指标来替代这些 Pod。
136+
137+
<!--
138+
## What's new for the beta?
139+
140+
For Kubernetes v1.27, the ContainerResource type metric is available by default as described at the beginning
141+
of this article.
142+
(You can still disable it by the `HPAContainerMetrics` feature gate.)
143+
-->
144+
## Beta 版本有哪些新内容？
145+
146+
在 Kubernetes v1.27 中，正如本文开头所述，ContainerResource 类型指标默认可用。
147+
（你仍然可以通过 `HPAContainerMetrics` 特性门禁用它。）
148+
149+
<!--
150+
Also, we've improved the observability of HPA controller by exposing some metrics from the kube-controller-manager:
151+
- `metric_computation_total`: Number of metric computations.
152+
- `metric_computation_duration_seconds`: The time that the HPA controller takes to calculate one metric.
153+
- `reconciliations_total`: Number of reconciliation of HPA controller.
154+
- `reconciliation_duration_seconds`: The time that the HPA controller takes to reconcile a HPA object once.
155+
-->
156+
另外，我们已通过从 kube-controller-manager 中公开一些指标来改进 HPA 控制器的可观测性：
157+
158+
- `metric_computation_total`：指标计算的数量。
159+
- `metric_computation_duration_seconds`：HPA 控制器计算一个指标所需的时间。
160+
- `reconciliations_total`：HPA 控制器的协调次数。
161+
- `reconciliation_duration_seconds`：HPA 控制器协调一次 HPA 对象所需的时间。
162+
163+
<!--
164+
These metrics have labels `action` (`scale_up`, `scale_down`, `none`) and `error` (`spec`, `internal`, `none`).
165+
And, in addition to them, the first two metrics have the `metric_type` label
166+
which corresponds to `.spec.metrics[*].type` for a HorizontalPodAutoscaler.
167+
-->
168+
这些指标具有 `action`（`scale_up`、`scale_down`、`none`）和
169+
`error`（`spec`、`internal`、`none`）标签。
170+
除此之外，前两个指标还具有 `metric_type` 标签，该标签对应于
171+
HorizontalPodAutoscaler 的 `.spec.metrics[*].type`。
172+
173+
<!--
174+
All metrics are useful for general monitoring of HPA controller,
175+
you can get deeper insight into which part has a problem, where it takes time, how much scaling tends to happen at which time on your cluster etc.
176+
-->
177+
所有指标都可用于 HPA 控制器的常规监控，你可以深入洞察哪部分存在问题，在哪里耗时，
178+
集群在哪个时间倾向于发生多少次扩缩等问题。
179+
180+
<!--
181+
Another minor stuff, we've changed the `SuccessfulRescale` event's messages
182+
so that everyone can check whether the events came from the resource metric or
183+
the container resource metric (See [the related PR](https://github.com/kubernetes/kubernetes/pull/116045)).
184+
-->
185+
另一件小事是，我们已更改了 `SuccessfulRescale` 事件的消息，
186+
这样每个人都可以检查事件是否来自资源指标或容器资源指标
187+
（请参见[相关 PR](https://github.com/kubernetes/kubernetes/pull/116045)）。
188+
189+
<!--
190+
## Getting involved
191+
192+
This feature is managed by [SIG Autoscaling](https://github.com/kubernetes/community/tree/master/sig-autoscaling).
193+
Please join us and share your feedback. We look forward to hearing from you!
194+
-->
195+
## 参与其中
196+
197+
此特性由 [SIG Autoscaling](https://github.com/kubernetes/community/tree/master/sig-autoscaling)
198+
进行管理。请加入我们分享反馈。我们期待聆听你的声音！
199+
200+
<!--
201+
## How can I learn more?
202+
203+
- [The official document of the ContainerResource type metric](/docs/tasks/run-application/horizontal-pod-autoscale/#container-resource-metrics)
204+
- [KEP-1610: Container Resource based Autoscaling](https://github.com/kubernetes/enhancements/tree/master/keps/sig-autoscaling/1610-container-resource-autoscaling)
205+
-->
206+
## 了解更多
207+
208+
- [ContainerResource 类型指标的正式文档](/zh-cn/docs/tasks/run-application/horizontal-pod-autoscale/#container-resource-metrics)
209+
- [KEP-1610：Container Resource based Autoscaling（基于容器资源的自动扩缩）](https://github.com/kubernetes/enhancements/tree/master/keps/sig-autoscaling/1610-container-resource-autoscaling)

0 commit comments

Comments
 (0)