nailgun-agent log too verbose

[Mike Scherbakov]

Nailgun agent logs logs too much data. In 1.5h it generated 3.3Mb file for 1 node.
It means that in 24h it's going to be 52.8Mb, and for average size cluster, let's say 25 nodes = 1320Mb in 24 hours.

This is way too much, and we can't handle to keep that much of data for diagnostic snapshot.

Let's optimize it. There are might be options how to do it, please find the best easy & possible (so not spend days on this problem):
1) log verbose only on first agent run on node
2) log only changes (looks like way more complicated)
3) disable such verbosity at all.

[Dima Shulyak (dshulyak)]

I like 1st option, with some modification:

If uid already received - log level should be INFO, otherwise DEBUG

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/135294

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/135294
Committed: https://git.openstack.org/cgit/stackforge/fuel-web/commit/?id=3366486fdb8d283563419d539d7faacc6ce3c127
Submitter: Jenkins
Branch: master

commit 3366486fdb8d283563419d539d7faacc6ce3c127
Author: Dima Shulyak <email address hidden>
Date: Tue Nov 18 16:08:41 2014 +0200

    Decrese agent to log to INFO level after uid is received

    Biggest part of agent logs is information about disks discovery
    which is DEBUG level

    Any changes with disks will be available in nailgun api log

    Change-Id: Idecbb1a4b045ea126acd04b6164b93e9c1fb1d0f
    Closes-Bug: 1393115