Am Fri, 29 Dec 2017 22:56:59 -0500
schrieb Nick Holland <n...@holland-consulting.net>:

> On 12/29/17 12:00, Michael Hekeler wrote:
> ...
> > I want to keep track of my changings in configfiles
> > like "/etc/ssh/sshd_config" or "~/.tmux.conf" and so
> > 
> > Normally I create "/root/RCS" and "~/RCS".
> > Then in every directory with configfiles that I want to change I
> > create a symlink ./RCS -> /root/RCS (in the example of sshd_config I
> > will create /etc/ssh/RCS as symlink to /root/RCS. So when I check in
> > sshd_config the revision file goes to /root/RCS
> > When I setup a new machine I can look in the older host's /root/RCS
> > and it shows me which files I have to edit (or better: which files
> > I edited on that host).
> > 
> > I am sure that every admin has its own way to do that. But I know
> > that it is always a good idea to listen carefully to more
> > experienced people.
> > That´s why I am asking.
> 
> One thing I have done for years, since hard disks became too stupidly
> big to even dream of using all of in many cases, is carve out a
> partition that I store dated tar files of the /etc/ directory in.
> 
> So -- /bu/etc20171220.tgz   /bu/etc20171221.tgz,
> and so on.  With compression, you can get YEARS of backup files in a
> 40g partition.
> 
> No check in/check out.  Diffing is non-trivial, but ... how often do
> you do it?  If you knew it worked yesterday(/last week) and is broke
> today, restore yesterday(/last week)'s files and figure out why it
> broke after you are back up and running.
> 
> For files like DNS zone files and pf.conf files, I wrote a script that
> you run on either machine, it shows you the diff to the other machine,
> has you comment/explain your change, then pushes your change over to
> the other file.  Works great for things where you have two different
> machines that should normally be running the same data, but you need
> to change and test that data from time to time.  Done properly, you
> get everything good from "change control" and revision control, and
> almost zero effort on the part of the administrators.  (yes, in the
> case of DNS, it means you handle the replication manually rather than
> through zone transfers -- and handling it manually is much better
> than the idiotic DNS master/slave concept.  Win all around).
> 
> Nick.
> 

Interesting :-)
 - thank you.

-- 
Viele Grüße
Michael

Reply via email to