H3Cloud OS集群环境安装显示安装包数量不一致问题处理经验案例

一线在客户现场于物理机服务器安装H3Cloud OS E1137H10版本的集群环境时，发现两个Cluster节点安装的文件包数量不一致。 一个Cluster节点显示安装包为969个，另一个Cluster是973个。

而两台服务器的配置及型号一致，具体配置信息如下表所示：

完成安装后，使用正常。

不涉及。

H3Cloud OS安装程序会判断当前服务器硬件配置的，有些包是和驱动相关的，不同配置服务器装的都不一样。另外服务器BIOS的设置会影响部门驱动程序。对于已完成安装的环境，可以在两个Cluster节点上执行命令“rpm -qa > /home/output.txt”将文件下载对比分析，具体安装文件包的区别。

对比具体文件的区别时，普通的文件对比工具是根据字符的位置来对比两个文件，而在分析过程中发现，安装文件包的先后顺序可能根据服务器的配置不同而不同，因此下面单独给出了一个Linux系统下的Shell脚本来对比文件。

Shell脚本的具体内容如下：

#!/usr/bin/bash

### compare two file content

###usage： ./twoFileCmp.sh  file1 file2

while read line

do

    #echo "File:  ${line}"

    grep "${line}*" $2 >/tmp/tmpfile

    linesNum=`wc -l /tmp/tmpfile | awk &＃39;{print $1}&＃39;`

    if [ $linesNum -eq 0 ];then

       echo "${line} ^~^ is not in $2"

    fi

    rm /tmp/tmpfile

done < $1

具体执行对比分析如下：

[root@cloudos fileComp]# ./twoFileCmp.sh  output_cluster227.txt  output_cluster228.txt

BDB2053 Freeing read locks for locker 0x34: 15990/140707951572992 ^~^ is not in output_cluster228.txt

BDB2053 Freeing read locks for locker 0x36: 15990/140707951572992 ^~^ is not in output_cluster228.txt

BDB2053 Freeing read locks for locker 0x37: 15990/140707951572992 ^~^ is not in output_cluster228.txt

BDB2053 Freeing read locks for locker 0x38: 15990/140707951572992 ^~^ is not in output_cluster228.txt

BDB2053 Freeing read locks for locker 0x39: 15990/140707951572992 ^~^ is not in output_cluster228.txt

BDB2053 Freeing read locks for locker 0x3a: 15990/140707951572992 ^~^ is not in output_cluster228.txt

BDB2053 Freeing read locks for locker 0x3b: 15990/140707951572992 ^~^ is not in output_cluster228.txt

BDB2053 Freeing read locks for locker 0x3c: 15990/140707951572992 ^~^ is not in output_cluster228.txt

BDB2053 Freeing read locks for locker 0x3d: 15990/140707951572992 ^~^ is not in output_cluster228.txt

BDB2053 Freeing read locks for locker 0x3e: 15990/140707951572992 ^~^ is not in output_cluster228.txt

BDB2053 Freeing read locks for locker 0x3f: 15990/140707951572992 ^~^ is not in output_cluster228.txt

BDB2053 Freeing read locks for locker 0x40: 15990/140707951572992 ^~^ is not in output_cluster228.txt

BDB2053 Freeing read locks for locker 0x41: 15990/140707951572992 ^~^ is not in output_cluster228.txt

grub2-2.02-0.44.el7.centos.x86_64 ^~^ is not in output_cluster228.txt

 [root@cloudos fileComp]#

[root@cloudos fileComp]# ./twoFileCmp.sh  output_cluster228.txt  output_cluster227.txt

BDB2053 Freeing read locks for locker 0x3f: 16016/140614638761984 ^~^ is not in output_cluster227.txt

BDB2053 Freeing read locks for locker 0x41: 16016/140614638761984 ^~^ is not in output_cluster227.txt

BDB2053 Freeing read locks for locker 0x42: 16016/140614638761984 ^~^ is not in output_cluster227.txt

BDB2053 Freeing read locks for locker 0x43: 16016/140614638761984 ^~^ is not in output_cluster227.txt

BDB2053 Freeing read locks for locker 0x44: 16016/140614638761984 ^~^ is not in output_cluster227.txt

BDB2053 Freeing read locks for locker 0x45: 16016/140614638761984 ^~^ is not in output_cluster227.txt

BDB2053 Freeing read locks for locker 0x46: 16016/140614638761984 ^~^ is not in output_cluster227.txt

BDB2053 Freeing read locks for locker 0x47: 16016/140614638761984 ^~^ is not in output_cluster227.txt

BDB2053 Freeing read locks for locker 0x48: 16016/140614638761984 ^~^ is not in output_cluster227.txt

BDB2053 Freeing read locks for locker 0x49: 16016/140614638761984 ^~^ is not in output_cluster227.txt

BDB2053 Freeing read locks for locker 0x4a: 16016/140614638761984 ^~^ is not in output_cluster227.txt

BDB2053 Freeing read locks for locker 0x4b: 16016/140614638761984 ^~^ is not in output_cluster227.txt

BDB2053 Freeing read locks for locker 0x4c: 16016/140614638761984 ^~^ is not in output_cluster227.txt

grub2-efi-2.02-0.44.el7.centos.x86_64 ^~^ is not in output_cluster227.txt

mokutil-0.9-2.el7.x86_64 ^~^ is not in output_cluster227.txt

efibootmgr-0.8.0-10.el7.x86_64 ^~^ is not in output_cluster227.txt

efivar-libs-0.11-1.el7.x86_64 ^~^ is not in output_cluster227.txt

shim-0.9-2.el7.x86_64 ^~^ is not in output_cluster227.txt

[root@cloudos fileComp]#

根据上述输出发现不同的文件包安装情况如下：

1、客户的Cluster1（IP以227结尾的服务器）中安装不同于Cluster2（IP以228结尾的服务器）文件包为：

grub2-2.02-0.44.el7.centos.x86_64 ^~^ is not in output_cluster228.txt

2、客户的Cluster2（IP以228结尾的服务器）中安装不同于Cluster1（IP以227结尾的服务器）文件包为：

grub2-efi-2.02-0.44.el7.centos.x86_64 ^~^ is not in output_cluster227.txt

mokutil-0.9-2.el7.x86_64 ^~^ is not in output_cluster227.txt

efibootmgr-0.8.0-10.el7.x86_64 ^~^ is not in output_cluster227.txt

efivar-libs-0.11-1.el7.x86_64 ^~^ is not in output_cluster227.txt

shim-0.9-2.el7.x86_64 ^~^ is not in output_cluster227.txt

综上反推发现，两节点安装时选择的引导模式不同，IP为228结尾的节点2选择的是UEFI的方式。

安装前检查服务器的引导模式设置，集群中节点配置相同的引导模式。

不涉及。