[et-mgmt-tools] Enhancements to cobbler replicate in git (Running multiple cobbler servers? This is for you)

[Michael DeHaan]

Hi folks,

I've added some things to the "cobbler replicate" command (thanks again 
to Scott Henson for coding that up).  Now, in addition to syncing 
metadata for distributions and profiles, it also does repos, and 
optionally can do systems as well.   It also has flags for being able to 
rsync tree and repo data, as well as the snippets and kickstart files 
that are used.    Other content like the settings file, that needs to be 
set up specifically for each server, is not included in "replicate".

So, why would you want this?  Suppose you want to manage a cobbler 
server at several sites to have them be as fast as possible.   However, 
you really don't want to have to make sure you send the same edits to 
every server -- you really just want to edit the items at the central 
server and have the remote cobbler servers clone them from that.

There are some limitations to this -- currently it will override objects 
with the copy from the master when you use this... there's no "merge" 
kind of functionality because we don't keep "last modified" data on each 
field.    Also, since we don't know if something from central is used 
remotely, we don't delete things that are removed on the central 
server.   However, you can make this happen by just deleting the 
/var/lib/cobbler/(distros|profiles|systems|repos) files on the remote 
prior to running the replicate command.   

The syntax is:

[root at mdehaan cobbler]# cobbler replicate --help
Usage: cobbler replicate  [ARGS]

Options:
  -h, --help         show this help message and exit
  --master=MASTER    Cobbler server to replicate from.
  --include-systems  include systems in addition to distros, profiles, and
                     repos
  --full-data-sync   rsync everything
  --sync-kickstarts  rsync kickstart templates
  --sync-trees       rsync imported trees
  --sync-triggers    rsync trigger scripts
  --sync-repos       rsync mirrored repo data

So, for example, to replicate as much as possible, it's just:

    cobbler replicate --master=centralserver.example.org 
--include-systems --full-data-sync

If you don't want to pull down the system records (I imagine this is 
usually true, as they are site specific)

    cobbler replicate --master=centralserver.example.org --full-data-sync

If you are replicating via some other means (Net App mirroring, etc), 
you can run your own update scrips first, and then do the following
to get the same features as before:

   cobbler replicate --master=centralserver.example.org

The additional options just provide some of the things that 
"--full-data-sync" provides, for intance, you might want to sync the 
repos and trees, but not the triggers, so that's why there are seperate 
options for those.

I'll put this up on the Wiki after getting some comments on what else 
might need to be added and whether this covers the potential use cases.

Thoughts on whether this solves the needs of maintaing multiple cobbler 
servers centrally?    Testing is welcome of course, this is in the 
development branch of git at the moment (1.1.X) and the code is 
relatively simple -- action_replicate.py.

Thanks!

--Michael