sqlalchemy GET /v1/resources returns too much meter data needs to match mongodb implementation

[John Tran]

The resources returned has meter attribute nested and those results are each and every row of the meters for that resource as well as all fields of each row. If a resource has tens of thousands of meters associated to it, the eager loading will load all of those rows and return all that data in the json result. Instead, it should match the mongodb implementation which simply returns a distinct list of unique {'counter_name': value, 'counter_type': value, 'counter_unit': value}.

[Julien Danjou (jdanjou)]

Fixing in https://review.openstack.org/#/c/23534/

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/23534
Committed: http://github.com/openstack/ceilometer/commit/98ac1d30f59f725cdbcb886501f08d3921a232a5
Submitter: Jenkins
Branch: master

commit 98ac1d30f59f725cdbcb886501f08d3921a232a5
Author: Julien Danjou <email address hidden>
Date: Mon Mar 4 18:33:39 2013 +0100

    api: run list_resources test against SQL

    This led me to discover and fix bug #1144837, which makes the SQL engine
    returns correctly the list of resources when asked.

    This also fixes a bug in the /resources listing in MongoDB driver: it used
    to filter on the resource db, which only contains the most recent up to date
    version of some fields, and returns wrong results.

    This also fixes bug #1131349

    Change-Id: I7441a8f2839c7db198c80de75f5c39f0dd219813
    Signed-off-by: Julien Danjou <email address hidden>