Greg,
I'm not so sure the dependency is really circular -- it's more like a
co-dependency, since Chef computes the list of cookbooks at startup time.
If we agree on that, it would be nicer, IMO, to replace some of the
logic in the PostgreSQL cookbook wrt setting up the "postgres" user
with a call to the user LWRP from the database cookbook.
- Julian
On Fri, Jan 4, 2013 at 3:29 PM, Greg Symons
<
<mailto: >>
wrote:
I thought I had responded to this the first time I saw it, but
I've got some preliminary work on 2 at
http://github.com/__DrillingInfo/postgresql
<http://github.com/DrillingInfo/postgresql> on the 'di' branch,
but I just came up with that stuff based on reading the docs, so
if you've got something better, go ahead and ignore it. That work
is _not_ compatible with the latest postgresql cookbook, though;
it's on my todo list to make it so.
Enhancement (3) should probably be a separate cookbook that
depends on both postgresql and database. Circular dependencies are
bad.
On Thu 03 Jan 2013 09:22:36 AM CST, David Crane wrote:
My current work project is to distill my postgresql
administration experience
into chef cookbooks. Here's my project plan ...
(1) I began (COOK-2051) by broadening cookbook support from
debian to redhat
platform families and from PostgreSQL-8.4 to PostgreSQL-9.X.
In this, I
continued the recent approach of using common files.
(2) Compute attribute defaults to tune postgresql.conf to
hardware resources
detected by ohai.
(3) Write recipes to manage postgresql group/login roles
through data_bags,
which will greatly help with multiple-server shops such as ours.
(4) Further work might provide support for replication from a
postgresql master
server to a hot standby.
Enhancement (3) leads to the architectural question: Would it
be OK to
introduce a dependency so that opscode-cookbooks/postgresql uses
opscode-cookbooks/database resources? It provides a
database_user resource that
I need to implement my group/role recipes.
Archive powered by MHonArc 2.6.16.