Yves Dorfsman:
>> The sad part is that you consider this unorthodox.
>> RCS, mercurial, git, etc... are so cheap to use that there is absolutely no
>> reason not to use them for NIS, DNS, ldif files etc...

Jesse Becker:
> To take it a step further:  any machine should have its configuration
> management  tool rules/promises/recipes under VC.


I also agree that configuration files, rules/promises/recipes or other
directory lookup files aren't really "unorthodox" uses of VC.
Although I would definitely lean toward rules/promises/recipes over
importing /etc into local or one or more central repositories.  I
can't imagine /etc repositories scaling very well to thousands of
machines...whereas a unified collection of rules/promises/recipes
under a single VC repository certainly would scale with little
issue...at least from the VC side of things.

Something else I've done in the past which I don't necessarily
consider unorthodox:  putting websites under VC...particularly
"canned" web services such as RT, Wordpress, Gallery, etc.  Especially
if the source is checked out from the repository instead of deploying
it by tarball.  If I use a cloned git repository as an example, then
updating the underlying website code could be as straight forward as
re-basing your local configuration commits on top of the new HEAD of
the remote branch.  (With subversion I think the equivalent would be
'svn update' to update the underlying revision to the latest one.)

I guess another example of an unorthodox use would be, as Adam Compton
suggested, using a centralized VC repository to synchronize/share
files between multiple hosts.  A similar idea I've been tossing around
for an environment is to maintain a site-wide Perl library under VC.
This would be a mix of CPAN and in-house modules that aren't included
in the core nor in the base install from the vendor.  My reasoning is
that CPAN (or CPANPlus, though I haven't extensively used it yet) does
a pretty good job of resolving it's dependencies, but it doesn't scale
well for keeping multiple hosts in sync.  At the same time, building
rpms for Perl modules, perhaps via cpan2rpm, is suboptimal because of
the dependencies that get left out.  Not only that, but the rpms that
are deployed are then disconnected from any documentation on _why_
they were deployed.

Obviously, the Perl library would then be kept in sync across multiple
hosts by "checking out" or "cloning" the library on them at regular
intervals.

--Aaron
_______________________________________________
Discuss mailing list
[email protected]
https://lists.lopsa.org/cgi-bin/mailman/listinfo/discuss
This list provided by the League of Professional System Administrators
 http://lopsa.org/

Reply via email to