Skip to content

Commit a270a64

Browse files
k8s-ci-robotk8s-ci-robot
authored
Merge pull request #20475 from ydcool/schedule-eviction-zh
Rename "Scheduling" concept as "Scheduling and Eviction" for zh
2 parents a40b776 + 32fa93d commit a270a64

File tree

7 files changed

+556
-696
lines changed

7 files changed

+556
-696
lines changed

‎content/zh/docs/concepts/scheduling-eviction/_index.md

Lines changed: 4 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,4 @@
1+
---
2+
title: "调度和驱逐(Scheduling and Eviction)"
3+
weight: 90
4+
---

‎content/zh/docs/concepts/scheduling-eviction/kube-scheduler.md

Lines changed: 169 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,169 @@
1+
---
2+
title: Kubernetes 调度器
3+
content_template: templates/concept
4+
weight: 50
5+
---
6+
7+
<!--
8+
---
9+
title: Kubernetes Scheduler
10+
content_template: templates/concept
11+
weight: 50
12+
---
13+
-->
14+
{{% capture overview %}}
15+
16+
<!--
17+
In Kubernetes, _scheduling_ refers to making sure that {{< glossary_tooltip text="Pods" term_id="pod" >}}
18+
are matched to {{< glossary_tooltip text="Nodes" term_id="node" >}} so that
19+
{{< glossary_tooltip term_id="kubelet" >}} can run them.
20+
-->
21+
在 Kubernetes 中，_调度_ 是指将 {{< glossary_tooltip text="Pod" term_id="pod" >}} 放置到合适的
22+
{{< glossary_tooltip text="Node" term_id="node" >}} 上，然后对应 Node 上的 {{< glossary_tooltip term_id="kubelet" >}} 才能够运行这些 pod。
23+
24+
{{% /capture %}}
25+
26+
{{% capture body %}}
27+
<!--
28+
## Scheduling overview {#scheduling}
29+
-->
30+
## 调度概览 {#scheduling}
31+
32+
<!--
33+
A scheduler watches for newly created Pods that have no Node assigned. For
34+
every Pod that the scheduler discovers, the scheduler becomes responsible
35+
for finding the best Node for that Pod to run on. The scheduler reaches
36+
this placement decision taking into account the scheduling principles
37+
described below.
38+
-->
39+
调度器通过 kubernetes 的 watch 机制来发现集群中新创建且尚未被调度到 Node 上的 Pod。调度器会将发现的每一个未调度的 Pod 调度到一个合适的 Node 上来运行。调度器会依据下文的调度原则来做出调度选择。
40+
41+
<!--
42+
If you want to understand why Pods are placed onto a particular Node,
43+
or if you're planning to implement a custom scheduler yourself, this
44+
page will help you learn about scheduling.
45+
-->
46+
如果你想要理解 Pod 为什么会被调度到特定的 Node 上，或者你想要尝试实现一个自定义的调度器，这篇文章将帮助你了解调度。
47+
48+
<!--
49+
## kube-scheduler
50+
-->
51+
## kube-scheduler
52+
53+
<!--
54+
[kube-scheduler](https://kubernetes.io/docs/reference/command-line-tools-reference/kube-scheduler/)
55+
is the default scheduler for Kubernetes and runs as part of the
56+
{{< glossary_tooltip text="control plane" term_id="control-plane" >}}.
57+
kube-scheduler is designed so that, if you want and need to, you can
58+
write your own scheduling component and use that instead.
59+
-->
60+
[kube-scheduler](/zh/docs/reference/command-line-tools-reference/kube-scheduler/) 是 Kubernetes 集群的默认调度器，并且是集群 {{< glossary_tooltip text="控制面" term_id="control-plane" >}} 的一部分。如果你真的希望或者有这方面的需求，kube-scheduler 在设计上是允许你自己写一个调度组件并替换原有的 kube-scheduler。
61+
62+
<!--
63+
For every newly created pods or other unscheduled pods, kube-scheduler
64+
selects a optimal node for them to run on. However, every container in
65+
pods has different requirements for resources and every pod also has
66+
different requirements. Therefore, existing nodes need to be filtered
67+
according to the specific scheduling requirements.
68+
-->
69+
对每一个新创建的 Pod 或者是未被调度的 Pod，kube-scheduler 会选择一个最优的 Node 去运行这个 Pod。然而，Pod 内的每一个容器对资源都有不同的需求，而且 Pod 本身也有不同的资源需求。因此，Pod 在被调度到 Node 上之前，根据这些特定的资源调度需求，需要对集群中的 Node 进行一次过滤。
70+
71+
<!--
72+
In a cluster, Nodes that meet the scheduling requirements for a Pod
73+
are called _feasible_ nodes. If none of the nodes are suitable, the pod
74+
remains unscheduled until the scheduler is able to place it.
75+
-->
76+
在一个集群中，满足一个 Pod 调度请求的所有 Node 称之为 _可调度节点_。如果没有任何一个 Node 能满足 Pod 的资源请求，那么这个 Pod 将一直停留在未调度状态直到调度器能够找到合适的 Node。
77+
78+
<!--
79+
The scheduler finds feasible Nodes for a Pod and then runs a set of
80+
functions to score the feasible Nodes and picks a Node with the highest
81+
score among the feasible ones to run the Pod. The scheduler then notifies
82+
the API server about this decision in a process called _binding_.
83+
-->
84+
调度器先在集群中找到一个 Pod 的所有可调度节点，然后根据一系列函数对这些可调度节点打分，然后选出其中得分最高的 Node 来运行 Pod。之后，调度器将这个调度决定通知给 kube-apiserver，这个过程叫做 _绑定_。
85+
86+
<!--
87+
Factors that need taken into account for scheduling decisions include
88+
individual and collective resource requirements, hardware / software /
89+
policy constraints, affinity and anti-affinity specifications, data
90+
locality, inter-workload interference, and so on.
91+
-->
92+
在做调度决定时需要考虑的因素包括：单独和整体的资源请求、硬件/软件/策略限制、亲和以及反亲和要求、数据局域性、负载间的干扰等等。
93+
94+
<!--
95+
## Scheduling with kube-scheduler {#kube-scheduler-implementation}
96+
-->
97+
## kube-scheduler 调度流程 {#kube-scheduler-implementation}
98+
99+
<!--
100+
kube-scheduler selects a node for the pod in a 2-step operation:
101+
102+
1. Filtering
103+
2. Scoring
104+
-->
105+
kube-scheduler 给一个 pod 做调度选择包含两个步骤：
106+
107+
1. 过滤
108+
2. 打分
109+
110+
<!--
111+
The _filtering_ step finds the set of Nodes where it's feasible to
112+
schedule the Pod. For example, the PodFitsResources filter checks whether a
113+
candidate Node has enough available resource to meet a Pod's specific
114+
resource requests. After this step, the node list contains any suitable
115+
Nodes; often, there will be more than one. If the list is empty, that
116+
Pod isn't (yet) schedulable.
117+
-->
118+
过滤阶段会将所有满足 Pod 调度需求的 Node 选出来。例如，PodFitsResources 过滤函数会检查候选 Node 的可用资源能否满足 Pod 的资源请求。在过滤之后，得出一个 Node 列表，里面包含了所有可调度节点；通常情况下，这个 Node 列表包含不止一个 Node。如果这个列表是空的，代表这个 Pod 不可调度。
119+
120+
<!--
121+
In the _scoring_ step, the scheduler ranks the remaining nodes to choose
122+
the most suitable Pod placement. The scheduler assigns a score to each Node
123+
that survived filtering, basing this score on the active scoring rules.
124+
-->
125+
在打分阶段，调度器会为 Pod 从所有可调度节点中选取一个最合适的 Node。根据当前启用的打分规则，调度器会给每一个可调度节点进行打分。
126+
127+
<!--
128+
Finally, kube-scheduler assigns the Pod to the Node with the highest ranking.
129+
If there is more than one node with equal scores, kube-scheduler selects
130+
one of these at random.
131+
-->
132+
最后，kube-scheduler 会将 Pod 调度到得分最高的 Node 上。如果存在多个得分最高的 Node，kube-scheduler 会从中随机选取一个。
133+
134+
<!--
135+
There are two supported ways to configure the filtering and scoring behavior
136+
of the scheduler:
137+
-->
138+
支持以下两种方式配置调度器的过滤和打分行为：
139+
140+
<!--
141+
1. [Scheduling Policies](/docs/reference/scheduling/policies) allow you to
142+
configure _Predicates_ for filtering and _Priorities_ for scoring.
143+
1. [Scheduling Profiles](/docs/reference/scheduling/profiles) allow you to
144+
configure Plugins that implement different scheduling stages, including:
145+
`QueueSort`, `Filter`, `Score`, `Bind`, `Reserve`, `Permit`, and others. You
146+
can also configure the kube-scheduler to run different profiles.
147+
-->
148+
1. [调度策略](/docs/reference/scheduling/policies) 允许你配置过滤的 _谓词(Predicates)_ 和打分的 _优先级(Priorities)_ 。
149+
2. [调度配置](/docs/reference/scheduling/profiles) 允许你配置实现不同调度阶段的插件，包括：`QueueSort`, `Filter`, `Score`, `Bind`, `Reserve`, `Permit` 等等。你也可以配置 kube-scheduler 运行不同的配置文件。
150+
151+
{{% /capture %}}
152+
{{% capture whatsnext %}}
153+
154+
<!--
155+
* Read about [scheduler performance tuning](/docs/concepts/scheduling-eviction/scheduler-perf-tuning/)
156+
* Read about [Pod topology spread constraints](/docs/concepts/workloads/pods/pod-topology-spread-constraints/)
157+
* Read the [reference documentation](/docs/reference/command-line-tools-reference/kube-scheduler/) for kube-scheduler
158+
* Learn about [configuring multiple schedulers](/docs/tasks/administer-cluster/configure-multiple-schedulers/)
159+
* Learn about [topology management policies](/docs/tasks/administer-cluster/topology-manager/)
160+
* Learn about [Pod Overhead](/docs/concepts/configuration/pod-overhead/)
161+
-->
162+
163+
* 阅读关于 [调度器性能调优](/zh/docs/concepts/scheduling-eviction/scheduler-perf-tuning/)
164+
* 阅读关于 [Pod 拓扑分布约束](/zh/docs/concepts/workloads/pods/pod-topology-spread-constraints/)
165+
* 阅读关于 kube-scheduler 的 [参考文档](/zh/docs/reference/command-line-tools-reference/kube-scheduler/)
166+
* 了解关于 [配置多个调度器](/zh/docs/tasks/administer-cluster/configure-multiple-schedulers/) 的方式
167+
* 了解关于 [拓扑结构管理策略](/zh/docs/tasks/administer-cluster/topology-manager/)
168+
* 了解关于 [Pod 额外开销](/zh/docs/concepts/configuration/pod-overhead/)
169+
{{% /capture %}}

0 commit comments

Comments
 (0)