Clean up sample conf files

[Greg Lucas]

The trove sample conf files contain a number of settings that are specific to the test/trove-integration environment and should not be included when the product is installed. They are also inconsistent with the other OpenStack components in a number of ways.

Some specific issues to address:

 - remove the 20-char trove-integration passwords, e.g. rabbit_password = f7999d1955c5014aa32c
 - use SQLite (not MySQL) as the default database, like all the other services
 - don't set log dir/file, or default to /var/log/trove/xxx

[Greg Lucas (glucas-q)]

Looking at e.g. the keystone and cinder sample conf files: they leave all the options commented out. This makes the sample file more like documentation, showing what all the defaults are.

We could take a similar approach with trove where most/all settings are commented out and we use the cfg.py defaults.

[Nikhil Manchanda (slicknik)]

Greg: Some of the other OpenStack projects actually generate the sample.conf using tooling, based on the default config values in the python code (cfg.py). It would be awesome if we could do something similar for Trove.

[Greg Lucas (glucas-q)]

Nikhil - Agreed, that would be preferable. I'll take a look at how nova is doing that.

[Greg Lucas (glucas-q)]

Nikhil - Should we close this based on the changes that have already merged?

It may make sense to create a new bug for generating the sample conf files. That will require breaking up cfg.py or somehow annotating it to indicate which options go in which file.

[Greg Lucas (glucas-q)]

Ignore my previous comment, I see that a number of the proposed changes have not yet been made.

[Amrith Kumar (amrith)]

Updating importance to "Wishlist" (bugscrub, 2014-10-23)