On 20/06/2013 18:36, Anselm Lingnau wrote: > Alan McKinnon wrote: > >> These days rcs is primarily a sysadmin tool whereas CVS/SVN/Git are >> primarily developer's tools > > RCS sucks as a sysadmin tool, for two reasons: (a) it does not track a file's > permissions, and (b) in many cases several files have to be changed at the > same time to achieve some desired goal – even for very simple sysadmin tasks > like adding a user –, which RCS can't represent. (Even CVS and SVN can't > really do it; it takes a changeset-oriented VCS like TLA, Git or Mercurial to > get it right.) > > The proper thing to do, while we're discussing this, is to use something like > etckeeper [1] on top of Mercurial or Git. This has the added benefit of being > able to automatically push changes to a remote repository as they are > committed, i.e., instant configuration backup.
I'd submit that if you want that level of control you should be using puppet or chef. rcs shines for atomic changes to single files like /etc/aliases (that's how we use it here) > > [1] http://joeyh.name/code/etckeeper/ > >> But I'm not advocating we test rcs as we really don't need to do >> exhaustive testing. It's after all a statistical numbers game. > > I agree that this does not belong in a LPIC-1 or LPIC-2 exam. One might put > it > on the list for a hypothetical LPIC-3-level »best sysadmin practices« exam if > it wasn't as much of a contentious issue as it is. > > Anselm > -- Alan McKinnon [email protected] _______________________________________________ lpi-examdev mailing list [email protected] http://list.lpi.org/cgi-bin/mailman/listinfo/lpi-examdev
