db.service_update unused join to compute_node

[Boris Pavlovic]

Every time when we are updating service we are making join to compute_node.

We don't use compute_node data from returning value.

And this method is very offen called.
For example if we have about 1000 nodes there will be about 1k services. And each 10 seconds they are updated. So it produces about 100 req/sec.

So it is important to make this method faster.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/24849

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/24849
Committed: http://github.com/openstack/nova/commit/2b7caba1053f53251074ed1500bf5a8b55c9a78d
Submitter: Jenkins
Branch: master

commit 2b7caba1053f53251074ed1500bf5a8b55c9a78d
Author: Boris Pavlovic <email address hidden>
Date: Wed Mar 20 01:33:00 2013 +0400

    Refactor db.service_destroy and db.service_update methods

    Remove session parameter from db.service_get method

    Optimize db.service_destroy:
    Was:
    1 request with join to get service with compute_node
    1 request to delete service
    1(0) to delete compute_node
    Now:
    1 request (without join) to delete service
    1 request to delete compute_node

    Optimize db.service_update:
    Remove unused join.

    Fixes bug 1157442

    blueprint db-session-cleanup

    Change-Id: Ifcc8fdebfc95daec736464cbc9e16301116540aa