On 2011-09-01 07:08, Andrew Hume wrote:
i am building a distributed cluster application which now has
two distinct sets of configuration info:
a) cluster-wide, shared by all components
b) server-specific, shared by components on a single server
for now, these are human-readable and editable text files
(NOT xml). the most likely changes to this information will be
A) tweaking of values for a new type of server (fair few over a day or so)
B) cloning server2's config for server3
my question is: how would you prefer to arrange these in the file-system?
the two obvious answers are
1) a single file (say config.all) with internal stanzas labelled by server name
Of course not, I'm not even sure why you're asking!
2) two files, say config.clust and config.local. in this case, the install
process
would need to take the appropriate ref copy of config.server1 and install it as
config.local on server1.
Do you (or will you eventually) have just a few configuration parameters, or
would it make sense to split this further into several files (say one per
application)?
If there is a potential that one file won't be enough or become inconvenient,
create one directory per node / class.
Do the file record how the machine should work, or, what's been installed on it?
If you record anything historical, then I'd create one directory per node.
Will your nodes be different per node, or per class?
If two nodes of the same class will always be identical, and assuming you
don't record any historical information in there, then go by class.
--
Yves. http://www.SollerS.ca/
http://ipv6.SollerS.ca
http://blog.zioup.org/
_______________________________________________
Tech mailing list
[email protected]
https://lists.lopsa.org/cgi-bin/mailman/listinfo/tech
This list provided by the League of Professional System Administrators
http://lopsa.org/
