Re: [BackupPC-devel] Windows ACL backup

[Craig B. <cba...@us...>]

Thomas Temp  writes:

> I've been working on Windows ACL support for BackupPC.
> Here's my approach :

Thanks for looking into this.  One problem is that I have found
rsync/ssh/cygwin to be generally unreliable, so I use only
rsyncd/cygwin.  I wasn't aware of setacl on SF.

My long-term hope would be to modify rsync to fetch the ACLs
and send them.  Some people have posted some patches to rsync
to make this possible.

The idea is that BackupPC would not translate the ACLs - the ACL
data would be kept in opaque form in the attrib file (just like
the other file modes).  On a restore, BackupPC would send the
exact raw data back to the remote rsync.

>  - it is not well integrated in the BackupCP code. I inserted code in
>  BackupPC_dump and BackupPC_restore, but a new module could be more
>  appropriate. I am not satisfied with my choice of config parameters.
>  Variable expansion in config parameters is ad-hoc, and
>  there could be some issues if the values interfere with regexp
>  syntax --I believe there is alread expansion code hidden somewhere in
>  BackupPC. Also, my Perl looks like I was writing Python code;)
> 
>  - its output is not logged. I watch it from the command-line. So, no
>  error reporting from the user's point-of-view.

If you look at the end of BackupPC_dump you will see a function
UserCommandRun().  This shows an example of using $bpc->cmdVarSubstitute
to do variable substitution and shell escaping, and $bpc->cmdSystemOrEval
to actually run the command writing the output to the log file.

> I would also like comments by people who are familiar with Windows ACLs.

Yes, I would too.  I really don't understand Windows ACLs and the
security model very well.

> The Free Software offering for backup is strong and rich, but Windows
> ACL backup is still mostly missing. Not for long, I hope.

I agree.

Craig