Browse Source

update (#15027)

3.2.1-prepare
旺阳 1 year ago committed by GitHub
parent
commit
a9a621fdee
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
  1. 25
      docs/docs/en/guide/installation/kubernetes.md
  2. 25
      docs/docs/zh/guide/installation/kubernetes.md

25
docs/docs/en/guide/installation/kubernetes.md

@ -514,6 +514,31 @@ common:
SW_GRPC_LOG_SERVER_PORT: "11800"
```
### How to deploy specific components separately?
Modify the `api.enabled`, `alert.enabled`, `master.enabled`, or `worker.enabled` configuration items in the `values.yaml` file.
For example, if you need to deploy worker to both CPU and GPU servers in a cluster, and the worker uses different images, you can do the following:
```bash
# Install master, api-server, alert-server, and other default components, but do not install worker
helm install dolphinscheduler . --set worker.enabled=false
# Disable the installation of other components, only install worker, use the self-built CPU image, deploy to CPU servers with the `x86` label through nodeselector, and use zookeeper as the external registry center
helm install dolphinscheduler-cpu-worker . \
--set minio.enabled=false --set postgresql.enabled=false --set zookeeper.enabled=false \
--set master.enabled=false --set api.enabled=false --set alert.enabled=false \
--set worker.enabled=true --set image.tag=latest-cpu --set worker.nodeSelector.cpu="x86" \
--set externalRegistry.registryPluginName=zookeeper --set externalRegistry.registryServers=dolphinscheduler-zookeeper:2181
# Disable the installation of other components, only install worker, use the self-built GPU image, deploy to GPU servers with the `a100` label through nodeselector, and use zookeeper as the external registry center
helm install dolphinscheduler-gpu-worker . \
--set minio.enabled=false --set postgresql.enabled=false --set zookeeper.enabled=false \
--set master.enabled=false --set api.enabled=false --set alert.enabled=false \
--set worker.enabled=true --set image.tag=latest-gpu --set worker.nodeSelector.gpu="a100" \
--set externalRegistry.registryPluginName=zookeeper --set externalRegistry.registryServers=dolphinscheduler-zookeeper:2181
```
Please note that the above steps are for reference only, and specific operations need to be adjusted according to the actual situation.
## Appendix-Configuration
| Parameter | Description | Default |

25
docs/docs/zh/guide/installation/kubernetes.md

@ -513,6 +513,31 @@ common:
SW_GRPC_LOG_SERVER_PORT: "11800"
```
### 如何单独部署特定组件?
修改 `values.yaml` 文件中的 `api.enabled`, `alert.enabled` `master.enabled``worker.enabled` 配置项
例如,在一个集群中需要同时将 worker 部署到 CPU 服务器和 GPU 服务器,并且 worker 使用不同的镜像,可以这样做:
```bash
# 安装 master、api-server、alert-server以及其他默认组件,但是不安装 worker
helm install dolphinscheduler . --set worker.enabled=false
# 禁用其他组件的安装,只安装 worker,使用自行建构建的 CPU镜像,通过 nodeselector部署到附带 x86标签的 CPU服务器,使用 zookeeper作为外部注册中心
helm install dolphinscheduler-cpu-worker . \
--set minio.enabled=false --set postgresql.enabled=false --set zookeeper.enabled=false \
--set master.enabled=false --set api.enabled=false --set alert.enabled=false \
--set worker.enabled=true --set image.tag=latest-cpu --set worker.nodeSelector.cpu="x86" \
--set externalRegistry.registryPluginName=zookeeper --set externalRegistry.registryServers=dolphinscheduler-zookeeper:2181
# 禁用其他组件的安装,只安装 worker,使用自行建构建的 GPU 镜像,通过 nodeselector部署到附带 a100标签的 gpu服务器,使用zookeeper作为外部注册中心
helm install dolphinscheduler-gpu-worker . \
--set minio.enabled=false --set postgresql.enabled=false --set zookeeper.enabled=false \
--set master.enabled=false --set api.enabled=false --set alert.enabled=false \
--set worker.enabled=true --set image.tag=latest-gpu --set worker.nodeSelector.gpu="a100" \
--set externalRegistry.registryPluginName=zookeeper --set externalRegistry.registryServers=dolphinscheduler-zookeeper:2181
```
请注意,以上步骤仅供参考,具体操作需要根据实际情况进行调整。
## 附录-配置
| Parameter | Description | Default |

Loading…
Cancel
Save