[ 
https://issues.apache.org/jira/browse/VCL-1065?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16109123#comment-16109123
 ] 

ASF subversion and git services commented on VCL-1065:
------------------------------------------------------

Commit 1803661 from [email protected] in branch 'vcl/trunk'
[ https://svn.apache.org/r1803661 ]

VCL-1065
Reverted changes to update_cygwin.cmd previously committed which added an 
attempt to call mkgroup.exe and mkpasswd.exe with "-l localhost" arguments. 
Added sed.exe commands to strip the '<HOST>+ section from the beginning of 
lines in group and passwd if it exists.

Removed "-t rsa1" argument from ssh-keygen.exe command. Recent versions don't 
support this and display an error. Without the argument, the key will be 
generated with the default type.

Added command at the end of update_cygwin.cmd to delete the 'VCL Update Cygwin' 
scheduled task if it exists. It doesn't need to be called on every reboot.

Added call to Windows.pm::post_load to delete the 'VCL Post Load' scheduled 
task if it exists. Also added identical call to 
Windows.pm::create_update_cygwin_startup_scheduled_task to make sure that both 
the 'VCL Post Load' and 'VCL Update Cygwin' scheduled tasks don't both exist 
because they cause conflicts.

Added call to 8.pm::pre_capture to delete the 'VCL Update Cygwin' scheduled 
task if it exists before adding the 'VCL Post Load' scheduled task.

Added '/RL HIGHEST' argument to schtasks.exe call in 
Windows.pm::create_startup_scheduled_task. It causes the task to run with 
"highest privileges". This seems to have corrected issues on at least one image 
where update_cygwin.cmd failed to run because of what seems to be a UAC issue.

Added call to Windows.pm::create_update_cygwin_startup_scheduled_task to copy 
the update_cygwin.cmd file from the management node to ensure the most recent 
copy is used.

> update_cygwin.cmd script does not work correctly if computer is joined to 
> Active Directory
> ------------------------------------------------------------------------------------------
>
>                 Key: VCL-1065
>                 URL: https://issues.apache.org/jira/browse/VCL-1065
>             Project: VCL
>          Issue Type: Sub-task
>          Components: vcld (backend)
>            Reporter: Andy Kurth
>            Assignee: Andy Kurth
>             Fix For: 2.5
>
>
> The GID and UID numbers Cygwin uses may change after a computer is joined to 
> Active Directory.
> Also, depending on the version of Cygwin the format of the resulting output 
> from *mkpasswd -l* and *mkgroup -l* may differ.  As a result, the operations 
> performed by *update_cygwin.cmd* may fail to allow the sshd service to work 
> correctly.  Example:
> Cygwin 1.5, output the same whether or not computer is joined to AD:
> {noformat}root:unused_by_nt/2000/xp:1000:513:U-VCLV98-247\root,S-1-5-21-4145529145-2666226658-940326903-1000:/home/root:/bin/bash{noformat}
> Note: the UID 1000 corresponds to the Windows numeric identifier for the user 
> account.  GID 513 represents the _None_ group.
> Cygwin 2.2, computer *not* joined to AD:
> {noformat}root:*:197609:197152:U-WIN10ENT\root,S-1-5-21-57237294-3426833820-2376726728-1001:/hom
>   e/root:/bin/bash{noformat}
> Note: the UID and GID numbers changed.  AFAIK, the _197152_ GID is fixed and 
> still corresponds to the _None_ group.
> Cygwin 2.8, computer *joined* to AD:
> {noformat}VCLV98-249+root:*:197609:197152:U-VCLV98-249\root,S-1-5-21-57237294-3426833820-2376726728-1001:/home/root:/bin/bash{noformat}
> Note: the computer name is prepended to the beginning of the line.  This is 
> also the case for the line corresponding to the _None_ group produced by the 
> mkgroup command:
> {noformat}
> Users:S-1-5-32-545:545:
> VCLV98-249+None:S-1-5-21-57237294-3426833820-2376726728-513:197121:
> {noformat}
> I also noticed problems caused by the way the vcld backend code configures 
> computers prior to a reboot or capture.  Each may add a startup scheduled 
> task -- _VCL Update Cygwin_ and _VCL Post Load_ respectively.  These may run 
> at the same time.  Both call update_cygwin.cmd (post load does a few 
> additional steps).  Problems occur if multiple instances of update_cygwin.cmd 
> runs at the same time.  Contained in the script are commands to delete 
> existing host key files and then regenerate them.  The regenerate command 
> hangs if the key file it is trying to create already exists.  This causes 
> that instance of update_cygwin.cmd to hang indefinitely.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

Reply via email to