Skip to main content
版本:预览版

flink-kubernetes-operator

Apache Flink (https://github.com/apache/flink-kubernetes-operator) 的 Kubernetes operator,它允许用户通过 kubectl 等原生 k8s 工具来管理 Flink 应用程序及其生命周期。

安装插件

vela addon enable flink-kubernetes-operator

卸载插件

vela addon disable flink-kubernetes-operator

由于这个插件依赖于fluxcdcert-manager插件,所以会自动启用它们。执行下面的命令来检查它们的状态:

$ vela ls -n vela-system
APP COMPONENT TYPE TRAITS PHASE HEALTHY STATUS CREATED-TIME
addon-cert-manager cert-manager helm running healthy Fetch repository successfully, Create helm release 2022-06-16 11:50:19 +0800 CST
successfully
addon-flink-kubernetes-operator flink-namespace raw running healthy 2022-06-16 11:50:20 +0800 CST
└─ flink-operator helm running healthy Fetch repository successfully, Create helm release 2022-06-16 11:50:20 +0800 CST
successfully
addon-fluxcd flux-system-namespace raw running healthy 2022-06-16 11:47:07 +0800 CST
└─ fluxcd-resources k8s-objects running healthy 2022-06-16 11:47:07 +0800 CST

通过显示flink-cluster组件的字段类型,让我们知道如何在一个应用程序中使用它们。 作为 flink 用户,你可以选择它们作为你的 flink 集群的设置参数。

vela show flink-cluster
# Properties
+--------------+-------------+--------+----------+---------------------------------------------------------------+
| NAME | DESCRIPTION | TYPE | REQUIRED | DEFAULT |
+--------------+-------------+--------+----------+---------------------------------------------------------------+
| name | | string | true | |
| namespace | | string | true | |
| nots | | string | true | 2 |
| flinkVersion | | string | true | v1_14 |
| image | | string | true | flink:latest |
| jarURI | | string | true | local:///opt/flink/examples/streaming/StateMachineExample.jar |
| parallelism | | int | true | 2 |
| upgradeMode | | string | true | stateless |
| replicas | | int | true | 1 |
| jmcpu | | int | true | 1 |
| jmmem | | string | true | 1024m |
| tmcpu | | int | true | 1 |
| tmmem | | string | true | 1024m |
+--------------+-------------+--------+----------+---------------------------------------------------------------+

首先请确保你的集群已经存在命名空间flink-home

然后部署下面的应用:

cat <<EOF | vela up -f -
apiVersion: core.oam.dev/v1beta1
kind: Application
metadata:
name: flink-app-v1
spec:
components:
- name: my-flink-component
type: flink-cluster
properties:
name: my-flink-cluster
namespace: flink-home
EOF

检查相关资源的状态:

vela status flink-app-v1 
About:

Name: flink-app-v1
Created at: 2022-04-22 17:33:51 +0800 CST
Status: running

Workflow:

mode: DAG
finished: true
Suspend: false
Terminated: false
Steps
- id:n6na24x6dr
name:my-flink-component
type:apply-component
phase:succeeded
message:

Services:

- Name: my-flink-component
Cluster: local Namespace: default
Type: flink-cluster
Healthy
No trait applied

你会看到你的第一个 flink-cluster 应用已经处于运行状态了。