Comment 1 for bug 1534303

Reviewed: https://review.openstack.org/267273
Committed: https://git.openstack.org/cgit/openstack/swift/commit/?id=3c0cf549f1e822cce8f905b069b317e676cf306b
Submitter: Jenkins
Branch: master

commit 3c0cf549f1e822cce8f905b069b317e676cf306b
Author: Samuel Merritt <email address hidden>
Date: Wed Jan 13 18:08:45 2016 -0800

    Speed up get_more_nodes() when there is an empty zone

    The ring has some optimizations in get_more_nodes() so that it can
    find handoffs that span all the regions/zones/et cetera and then stop
    looking. The stopping is the important part.

    Previously, it would quickly find a handoff in each unused region,
    then spend way too long looking for more unused regions; the same was
    true for zones, IPs, and so on. Thus, in commit 9cd7c6c, we started
    counting regions and zones, then stopping when we found them all.

    This count included all regions and zones in the ring, regardless of
    whether or not there were actually any parts assigned or not. In rings
    with an empty region, i.e. a region for which there are only
    zero-weight devices, get_more_nodes() would be very slow.

    This commit ignores devices with no assigned partitions when counting
    regions, zones, and so forth, thus greatly speeding things up.

    The output of get_more_nodes() is unchanged. This is purely an
    optimization.

    Closes-Bug: 1534303

    Change-Id: I4a5c57205e87e1205d40fd5d9458d4114e524332