Dask的集群启动创建也很简单，有好几种方式，最简单的是采用官方提供dask-scheduler和dask-worker命令行方式。本文描述如何使用命令行方法建立Dask集群。

一、概述

官方介绍如下：

这是在多台计算机上部署Dask的最基本方法。在生产环境中，此过程通常由其他资源管理器自动执行。因此，很少有人需要明确遵循这些说明。 相反，这些说明对可能想要设置自动化服务以在其机构内部署Dask的IT专业人员很有用。

主要步骤包括：
1、启动dask-scheduler
2、注册worker

假设我们有两台主机，一台主机作为scheduler，同时也作为worker，另一台主机只作为worker。使用方法如下：

一、启动scheduler

直接使用dask-scheduler即可。安装完dask之后，系统就会有dask-scheduler命令，在作为schedule的服务器上直接运行如下命令即可：

dask-scheduler

一般有如下结果表明创建scheduler成功：

$ dask-scheduler
Start scheduler at 127.0.0.1:8786

二、注册worker

与前面类似，直接使用dask-worker即可，同时加上scheduler的地址，然后使用默认配置启动了。

成功后有如下信息：

$ dask-worker 127.0.0.1:8786
Start worker at:               127.0.0.1:1234
Registered with scheduler at:  127.0.0.1:8786

当然worker中有很多配置，最常用的是线程数（nthreads）和进程数（nprocs），如下所示

$ dask-worker 127.0.0.1:8786 --nthreads 1 --nprocs 2
distributed.nanny - INFO -         Start Nanny at: 'tcp://127.0.0.1:37543'
distributed.nanny - INFO -         Start Nanny at: 'tcp://127.0.0.1:35610'
distributed.dashboard.proxy - INFO - To route to workers diagnostics web server please install jupyter-server-proxy: python -m pip install jupyter-server-proxy
distributed.dashboard.proxy - INFO - To route to workers diagnostics web server please install jupyter-server-proxy: python -m pip install jupyter-server-proxy
distributed.worker - INFO -       Start worker at: tcp://127.0.0.1:36678
distributed.worker - INFO -          Listening to: tcp://127.0.0.1:36678
distributed.worker - INFO -          dashboard at:       127.0.0.1:42342
distributed.worker - INFO -       Start worker at: tcp://127.0.0.1:46164
distributed.worker - INFO - Waiting to connect to:    tcp://127.0.0.1:8786
distributed.worker - INFO -          Listening to: tcp://127.0.0.1:46164
distributed.worker - INFO - -------------------------------------------------
distributed.worker - INFO -          dashboard at:       127.0.0.1:42046
distributed.worker - INFO -               Threads:                          1
distributed.worker - INFO - Waiting to connect to:    tcp://127.0.0.1:8786
distributed.worker - INFO -                Memory:                    4.19 GB
distributed.worker - INFO - -------------------------------------------------
distributed.worker - INFO -       Local Directory: /home/lodap/data/dask-worker-space/worker-a01ixxg6
distributed.worker - INFO - -------------------------------------------------
distributed.worker - INFO -               Threads:                          1
distributed.worker - INFO -                Memory:                    4.19 GB
distributed.worker - INFO -       Local Directory: /home/lodap/data/dask-worker-space/worker-ly7gp0_h
distributed.worker - INFO - -------------------------------------------------
distributed.worker - INFO -         Registered to:    tcp://127.0.0.1:8786
distributed.worker - INFO - -------------------------------------------------
distributed.core - INFO - Starting established connection