OpenSSH9.2导致cluvfy命令失败

1、案例概述

同事在AIX7.1下安装11.2.0.4 GI集群时，运行cluvfy命令失败。 报错信息如下：

Version of exectask could not be retrieved from node "skdb1"
ERROR:
Remote execution files could not be copied to "/tmp/CVU_11.2.0.4.0_grid" on the following nodes:
     skdb1
Verification will proceed with nodes:
     skdb2

 

2、故障处理过程

2.1 在执行cluvfy时，需要把当前节点/tmp/CVU_11.2.0.4.0_grid目录中的一些临时脚本复制至其他的节点并执行，如果/tmp目录的权限存在问题，则可能会出现这种情况。需要检查节点的/tmp目录情况。

mounted   mounted over    vfs       options
/dev/hd3    /tmp          jfs2     rw,log=/dev/hd8

可以看出，/tmp的权限是正常的。

2.2 此时，只能开启debug模式，获取更多的日志。例如：

[grid@db01~]$ rm -rf /tmp/cvutrace
[grid@db01~]$ mkdir /tmp/cvutrace
[grid@db01~]$ export CV_TRACELOC=/tmp/cvutrace
[grid@db01~]$ export SRVM_TRACE=true
[grid@db01~]$ export SRVM_TRACE_LEVEL=1
[grid@db01~]$ cluvfy ..... -verbose

2.3 分析生成的debug日志。

[2818456@nskdb1] [Thread-38] [ 2024-11-14 16:02:52.102 CST ] [StreamReader.run:65]  ERROR>/usr/bin/scp: dest open "'/tmp//CVU_11.2.0.4.0_grid/runfixup.sh'": No such file or directory
[2818456@nskdb1] [Thread-38] [ 2024-11-14 16:02:52.103 CST ] [StreamReader.run:65]  ERROR>/usr/bin/scp: failed to upload file /tmp/bootstrap/cv/remenv/runfixup.sh to '/tmp//CVU_11.2.0.4.0_grid/runfixup.sh'
[2818456@nskdb1] [Worker 0] [ 2024-11-14 16:02:52.106 CST ] [RuntimeExec.runCommand:144]  runCommand: process returns 1
[2818456@nskdb1] [Worker 0] [ 2024-11-14 16:02:52.106 CST ] [RuntimeExec.runCommand:161]  RunTimeExec: output>
[2818456@nskdb1] [Worker 0] [ 2024-11-14 16:02:52.106 CST ] [RuntimeExec.runCommand:170]  RunTimeExec: error>
[2818456@nskdb1] [Worker 0] [ 2024-11-14 16:02:52.106 CST ] [RuntimeExec.runCommand:173]  /usr/bin/scp: dest open "'/tmp//CVU_11.2.0.4.0_grid/runfixup.sh'": No such file or directory
[2818456@nskdb1] [Worker 0] [ 2024-11-14 16:02:52.106 CST ] [RuntimeExec.runCommand:173]  /usr/bin/scp: failed to upload file /tmp/bootstrap/cv/remenv/runfixup.sh to '/tmp//CVU_11.2.0.4.0_grid/runfixup.sh'
[2818456@nskdb1] [Worker 0] [ 2024-11-14 16:02:52.106 CST ] [RuntimeExec.runCommand:192]  Returning from RunTimeExec.runCommand

从debug日志可以看出，在执行scp命令将本地的runfixup.sh复制至远程主机时报错。

2.4 同事手动执行scp命令进行拷贝时，没有任何问题。 这一点非常奇怪，让同事检查了下SSH的版本信息。

# ssh -V
OpenSSH_9.2P1 OpenSSL 1.1.1x

2.5 当前SSH的版本为9.2, 怀疑是SSH版本过高，建议主机工程师降级SSH版本。 最终， SSH降到8.1版本后，问题解决。