counter_volume overflows in sqlalchemy implementation

[François Charlier]

In the SQLAlchemy implementation, the counter_volume column is an INT. As such the value range is from -2147483648 to 2147483647. Greater or lesser values are then truncated when inserted in the SQL database, resulting in dataloss.

This range is insufficient for many counters, as:
* cpu
* disk.read.bytes
* disk.write.bytes
* disk.write.requests
* image.download
* image.serve
* image.size
* storage.objects.size

I suggest to change the counter_volume to BIGINT.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/25006

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/25006
Committed: http://github.com/openstack/ceilometer/commit/d09b7b72267982c12b88a0eaa87a4ae4233f8a8d
Submitter: Jenkins
Branch: master

commit d09b7b72267982c12b88a0eaa87a4ae4233f8a8d
Author: François Charlier <email address hidden>
Date: Thu Mar 21 16:21:11 2013 +0100

    Change the column counter_volume to Float

    counter_volume was previously an Integer (32 bits value in SQL
    implementations as MySQL or PostgreSQL) and limited in values from
    -2147483648 to 2147483647, very small for things like disk I/O bytes,
    network I/O bytes, image size, object size, volume size, …
    Furthermore one might want to store some real numbers (ℝ) (temperatures,
    power comsumption, …)

    Change-Id: Ifdb8e8e84e8e01df267d71bc72cdc3f56bb2f695
    Fixes: bug #1158276