xenapi: resize operation could be faster

[John Garbutt]

There are a few small performance issues that can easily be fixed.

First a regression:
* on resize down we now call resize2fs twice, once to get the minimum file system size. There is no need to call it twice, just handle the error case correctly

The other issue:
* resize_disk uses copy, we could consider using clone (or snapshot) to save time, assuming that does not slow down the other steps
* consider using zerofree to help reduce the filesystem size further, after shrinking the filesystem, before the sparse copy
* while the above will take time, it should reduce the amount of data copied over the network

[Sahid Orentino (sahid-ferdjaoui)]

In the first point, I'm thinking about the kind of exception we have to raise:
     exception.ResizeError or processutils.ProcessExecutionError ?

[John Garbutt (johngarbutt)]

We already raise an exception here:
https://github.com/openstack/nova/blob/master/nova/virt/xenapi/vm_utils.py#L2069

We can just skip the min_blocks check, then deal with the particular return value from resize2fs that happens when the filesystem minimum size is too big.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/46531

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/46531
Committed: http://github.com/openstack/nova/commit/f246c3067b9608700dd29ee3828e094a9f0bc097
Submitter: Jenkins
Branch: master

commit f246c3067b9608700dd29ee3828e094a9f0bc097
Author: Sahid Orentino Ferdjaoui <email address hidden>
Date: Fri Sep 13 16:16:42 2013 +0000

    xenapi: Resize operations could be faster

     * Improves the code by using only one time resize2fs during the resize.
     * Removes the not used method: "_get_min_sectors".
     * Uses a VDI clone instead of copy in resize_disk's method.
     * Review of some test cases to accept the new behavior
       without _get_min_sectors.

    Closes-Bug: 1221332
    Change-Id: I40a1d7d467353ba904689d128de7f89d785b4052