I also attended the workshop and this is a good list. However, from
Tom's talk I took the idea that we need a canonical application of
configuration management so people can see it in action (e.g. like
google maps for ajax). I am thinking about some kind of demo where a
person can, for example, change a config file, and then watch as the
configuration management system change the file back (assuring
correctness). Another idea would be to migrate a web server from
machine A to machine B. This way we could show folks a tangible benefit
of configuration management.
Ski Kacoroski
--
"When we try to pick out anything by itself, we find it
connected to the entire universe" John Muir
Chris "Ski" Kacoroski, [EMAIL PROTECTED], 206-501-9803
Gulfie wrote:
On Tue, Dec 06, 2005 at 04:52:36PM -0600, Brendan Strejcek wrote:
Here is my initial very incomplete list, in no particular order.
- Keep /etc/issue in sync on a group of nodes
- All ssh known_hosts files should match the state of the whole fabric
- Maintain a site-wide scripts directory
- Hierarchical config file distribution
(Example: Bcfg Cfg generator, cfengine singlecopy)
- Configure node X to be a Y server and configure clients to match
- Make sure user X exists on all nodes with the same credentials
- Set a configuration parameter (in a config file, registry, ...)
and bring running services into line with the new policy
- /opt/gnu/bin/m4 should resolve to GNU m4 on all nodes
- Make sure /path/to/service/data is being backed up
- Make sure no processes with X characterstics are running
Additions:
- allocate resource, image, deploy and configure that resource to be a
server for service X
Where service can be simply a protocol or server name, or it can be more site specific like,
'one of the .jpg servers'.
- Migrate/replicate the data in repositor X from server A to server B
- Gracefully Decommision server X, scrub the data of it's disk, and
place it back into the free pool.
- Migrate service from target X to target Y. Example. NFS
1) install the new server/service
2) Majic + human interaction for downtime planning
3) done
- Duplicate 'this' inviroment over 'there'. (automatic DR/BC site
construction).
- Given parts on hand, construct a (whatever)
- Turn box X into a print server for the local lan/building. And have the rest of the
computational infrastructure deal with it well.
- Join two admin / user domains
- split two admin / user domains
- Given content field A, usage histograms B, and parts list C,
construct a service farm of Y
performance and Z reliability.
- Install apache on an allocated machine with mod perl and list of modules Y.
Note: the freedoms of OS, apache version, perl versions and module versions are left
as a freedom to the problem solver.
-gulfie
_______________________________________________
lssconf-discuss mailing list
lssconf-discuss@inf.ed.ac.uk
http://lists.inf.ed.ac.uk/mailman/listinfo/lssconf-discuss
