Hy all,
i have recently join the fedora infrastructure list, so maybe this survey would appear to you a bit odd.
i have taken a deeper look at what have been discussed regarding the config management task,
many of you have strong experience in different kind of tools like glump, puppet, cfengine ...
but i haven't seen any checkup list on :
- what's the aim of a new config management system.
- how far we want this tool to manage servers config (system,services, both of them, anything else maybe ....).
- system limitations (for example : no ruby tools because it has been sealed that ruby should be installed on servers ).
i've
seen jeff ollie has made a little wiki page listing all config
management tools suggested, with pros and cons,
here's my little suggestion:
first :
we should list :
- what the new config tool should do for us
- perhaps how we want this tool to work, (agent/server or something else)
- system things for example : no ruby, if there an agent so it should communicate securely with his server .....
second :
looking at jeff wiki's page and other solutions discussed on the list, and see if there is any tool listed there which answer all our requirements,
if not then 2 choice :
- take a look around and find the "right" tool
- lower a little bit our requirements, so one of the tools listed would answer them (definitely the worst solution ).
third:
-build a test environment with the chosen tool, so
everyone can play with it (the more people become familiar with it, the
best it would be ;) )
fourth:
- make a deployment schedule.
fifth:
- deploy and enjoy the work done by all people involved in ;=)
here my personal view:
1)
what the new config tool should do for us:
this tool should help us managing system and services configurations of all fp.o servers.
by managing, i mean :
- building new config to deploy.
- easily and smoothly updating several config
- versioning all config created and/or updated
how we want this tool to work:
- all tasks should be done securely, agent/server tool or not.
- the tool should give us some report on how tasks goes.
system things:
well as I'm not used on how servers are deployed , i can't really answer this point but,
IMHO, whatever tool is chosen, it should not require something that may change how our systems works, this tool should be nearly invisible system view.
2)
about the right tool, after looking at threads in the list archive and looking at jeff page, glump seems to be a good start
3) and 4) would come along the way after all question for 1) and 2) are settled.
who's next to answer this little survey ? :)
hoping this would help
Rachid
