Slurm 节点设置

本节介绍在 BCM 头节点上执行的配置步骤。

1. 首先，为 slogin 节点创建软件镜像和类别。

   cmsh
   
   [bcm10-headnode]% softwareimage
   [bcm10-headnode->softwareimage]% clone default-image slogin-image
   [bcm10-headnode->softwareimage*[slogin-image*]]% commit
   [bcm10-headnode->softwareimage[slogin-image]]% category
   [bcm10-headnode->category]% clone default slogin
   [bcm10-headnode->category*[slogin*]]% set softwareimage slogin-image
   [bcm10-headnode->category*[slogin*]]% commit
   

2. 现在创建第一个 slogin 节点。

   cmsh
   
   [bcm10-headnode]% device
   [bcm10-headnode->device]% clone node001 slogin-01
   [bcm10-headnode->device*[slogin-01*]]% set category slogin
   [bcm10-headnode->device*[slogin-01*]]% commit
   

3. 添加并配置接口。

   注意

   接口的名称将根据节点的硬件供应商而变化。

   cmsh
   
   [bcm10-headnode]% device
   [bcm10-headnode->device]% use slogin-01
   [bcm10-headnode->device[slogin-01]]% interfaces
   [bcm10-headnode->device[slogin-01]->interfaces]% add bmc ipmi0 10.133.10.21 ipminet
   [bcm10-headnode->device*[slogin-01*]->interfaces*[ipmi0*]]% commit
   [bcm10-headnode->device[slogin-01]->interfaces]% add physical enp65s0np0; add physical ens1f1np1
   [bcm10-headnode->device*[slogin-01*]->interfaces*[ens1f1np1*]]% commit
   [bcm10-headnode->device[slogin-01]->interfaces]% add bond bond0 10.133.11.21 internalnet
   [bcm10-headnode->device*[slogin-01*]->interfaces*[bond0*]]% append interfaces enp65s0np0 ens1f1np1
   [bcm10-headnode->device*[slogin-01*]->interfaces*[bond0*]]% remove bootif
   [bcm10-headnode->device*[slogin-01*]->interfaces*[bond0*]]% ..
   [bcm10-headnode->device*[slogin-01*]->interfaces*]% ..
   [bcm10-headnode->device*[slogin-01*]]% set provisioninginterface bond0
   [bcm10-headnode->device*[slogin-01*]]% commit
   

4. 克隆第一个 slogin 节点以添加其他管理节点。cpu-01 将用作其他管理节点的热备。

   cmsh
   
   [bcm10-headnode]% device
   [bcm10-headnode->device]% clone slogin-01 slogin-02 --next-ip
   [bcm10-headnode->device*[slogin-02*]]% commit
   [bcm10-headnode->device]% clone slogin-01 cpu-01 --next-ip
   [bcm10-headnode->device*[cpu-01*]]% set category default
   [bcm10-headnode->device*[cpu-01*]]% commit
   

5. 设置每个节点的 MAC 地址，以便它们可以 PXE 启动。

   对 slogin-02 和 cpu-01 重复此步骤

   cmsh
   [bcm10-headnode]% device
   [bcm10-headnode->device]% use slogin-01; interfaces
   [bcm10-headnode->device[slogin-01]->interfaces]% set enp65s0np0 mac 00:00:00:00:00:01
   [bcm10-headnode->device*[slogin-01*]->interfaces*]% set ens1f1np1 mac 00:00:00:00:00:02
   [bcm10-headnode->device*[slogin-01*]->interfaces*]% ..
   [bcm10-headnode->device*[slogin-01*]]% set mac 00:00:00:00:00:01
   [bcm10-headnode->device*[slogin-01*]]% commit
   

6. 开启电源并配置节点。