CONF.rescue_timeout not used correctly

[Andrew Laski]

The rescue_timeout should unrescue an instance after it has been in RESCUE mode for n seconds, but instead it serves to ensure unrescue logic is only run every n seconds. And when the unrescue logic is run all rescued instances are unrescued no matter how short of a time they have been in that state. At least with the Xen driver.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/17869

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/17869
Committed: http://github.com/openstack/nova/commit/c40fc8a4db3fe2d4d415f27d275b1d784a90cfe5
Submitter: Jenkins
Branch: master

commit c40fc8a4db3fe2d4d415f27d275b1d784a90cfe5
Author: Andrew Laski <email address hidden>
Date: Tue Dec 11 13:48:11 2012 -0500

    Fix poll_rescued_instances periodic task

    The poll_rescued_instances periodic task now checks the amount of time
    that an instance has been in the RESCUED stated before timing out the
    rescue. It also now performs the unrescue through the compute api in
    order to make sure the database is left in a consistent state.

    The poll_rescued_instances method is no longer necessary in the virt
    driver interface and has been removed. And also removed from the
    different virt drivers, since it was just doing a 'pass' in each of
    them.

    bug 1088625
    bug 1088627

    Change-Id: I75f7dc188cc49e5f6e5c8a3cb256d1c42ff7d882