Metro Node:通过无密码 SSH 连接到节点或控制器失败
Summary: 本文讨论了在尝试访问任何 Metro Node 节点/控制器时,系统会提示您输入密码的问题。这不是预期行为,因为在使用服务帐户时,Metro Node 支持在节点/控制器之间使用无密码 SSH。
Symptoms
受影响的 Metro Node 硬件:
戴尔硬件:Metro Node-mn114/mn215
Dell Hardware:Metro Node — Local 或 Metro
受影响的 Metro Node 操作系统版本:
戴尔软件:Metro Node OS 7.x
戴尔软件:Metro Node OS 8.x
问题:
尝试使用服务帐户通过 SSH 连接到任何 Metro Node 节点或控制器时,系统会提示您输入密码,如下所示:
示例:(下面的输出显示节点 B 无法执行到节点 A 的无密码 SSH)
service@director-1-1-b:~> ssh 128.221.252.35 Password:
症状:
所有控制器报告 director-x-y-z : Systemd IO error: Cannot determine the state of system 在“群集状态”输出中,如下所示:
示例:
VPlexcli:/> cluster status
WARNING: There are unreachable directors: director-1-1-A. Connectivity may still have errors even if none are reported.
Cluster cluster-1
operational-status: degraded
transitioning-indications: disk(s) not visible from all directors,meta data problem
transitioning-progress:
health-state: degraded
health-indications: director-1-1-A : Systemd IO error: Cannot determine the state of system services
local-com: connectivity: NONE
LC-00 ports - FAIL - Failed to determine expected connectivity.
LC-01 ports - FAIL - Failed to determine expected connectivity.
Cluster Name Port Group MTU Connectivity Status
------------ ---------- ---- ------------ -----------
cluster-1 LC-00 1500 fail all-enabled
LC-01 1500 fail all-enabled
cluster-2 LC-00 1500 fail all-enabled
LC-01 1500 fail all-enabled
man-com: connectivity: NONE
MC-01 ports - FAIL - Failed to determine expected connectivity.
MC-00 ports - FAIL - Failed to determine expected connectivity.
Cluster Name Port Group MTU Connectivity Status
------------ ---------- ---- ------------ -----------
cluster-1 MC-00 1500 fail all-enabled
MC-01 1500 fail all-enabled
cluster-2 MC-00 1500 fail all-enabled
MC-01 1500 fail all-enabled
Cluster cluster-2
operational-status: ok
transitioning-indications:
transitioning-progress:
health-state: ok
health-indications:
local-com: ok
man-com: ok
wan-com: ok
Cause
如果将服务目录权限修改为完全访问 (777)(读取、写入和运行权限),则可能会出现此问题
Resolution
解决方法步骤:
-
使用服务用户身份登录受影响的节点,然后通过在服务/Linux 提示符处输入以下命令,转至 service-maintenance-window:
sudo /usr/sbin/service-maintenance-window -f
-
检查并更改为受影响节点上的服务目录定义的权限,如下所示:
示例:(根据症状部分确定受影响的 Metro Node 节点 A)
From: director-1-1-a:/home # ll total 28 drwxr-x--- 6 admin users 4096 Aug 20 2021 admin drwxr-x--- 6 eseservice users 4096 Aug 20 2021 eseservice drwx------ 2 root root 16384 Mar 3 2022 lost+found drwxrwxrwx 11 service users 4096 May 20 05:08 service To: director-1-1-a:/home # chmod 750 service director-1-1-a:/home # ll total 28 drwxr-x--- 6 admin users 4096 Aug 20 2021 admin drwxr-x--- 6 eseservice users 4096 Aug 20 2021 eseservice drwx------ 2 root root 16384 Mar 3 2022 lost+found drwxr-x--- 11 service users 4096 May 20 05:08 service
-
尝试以服务用户身份通过 SSH 连接回受影响的节点,并确认您是否可以使用无密码 SSH,如下所示:
示例:(下面的输出显示,节点 B 可以成功执行到节点 A 的无密码 SSH)
service@director-1-1-b:~> ssh 128.221.252.35 Last login: Mon May 20 05:32:04 2024 from 10.107.104.132 service@director-1-1-a:~>
-
再次运行群集状态命令以确认未看到任何错误:
VPlexcli:/> cluster status Cluster cluster-1 operational-status: ok transitioning-indications: transitioning-progress: health-state: ok health-indications: local-com: ok man-com: ok Cluster cluster-2 operational-status: ok transitioning-indications: transitioning-progress: health-state: ok health-indications: local-com: ok man-com: ok wan-com: ok
Additional Information
*使用 chmod 绝对模式下
*在绝对模式下,权限以数字形式表示(准确地说是八进制)。在此系统中,每个文件权限都由一个数字表示。
r (读取) = 4
w (写入) = 2
x (执行) = 1
– (无权限) = 0
使用这些数值,您可以将它们组合在一起,因此可以使用一个数字来表示整个权限集。
| 编号 | 权限 |
|---|---|
| 0 | — |
| 1 | –X |
| 2 | -w- |
| 3(即 2+1) | -Wx |
| 4 | R– |
| 5(即 4+1) | R-X |
| 6(即 4+2) | 乌尔曼- |
| 7(即 4+2+1) | RWX系列 |