Open Stack

Ok, so part of the findings here is that when that flag is not enabled, it seems like the resize2fs and commands that run on the image file beforehand might actually screw up the image file (if it is qcow2). Has anyone tried this flag with a false value and used qcow2 images. Is there a patch to ubuntus resize2fs that makes it work with qcow2 images. I don't see how this could have worked without something like that previously (if ever).

It seems like the right way forward is to use libguestfs resize (which requires disk images to have a partition).

On 6/4/12 10:12 AM, "Joshua Harlow" <harlowja at yahoo-inc.com> wrote:

Hi all,

I was wondering if there has been anyone else who has used this flag with libguestfs (on RH 6.2) that has noticed file sync issues.

force_raw_images=true

I have been turning that to false so that images need not be expanded for actual usage (it seems this only affects the base images in _base) but when turning that to false and when file injection occurs, it seems that the files get written (I've added os.listdir and such for debugging) but when unmounted and later that image is started it seems like those files either do not get synced back to the image (I even added a manual 'sync' call) or there is some other corruption that happens that causes those images to not contain those files when started (for reference I am using qcow2 files).

Has anyone else seen this type of issue?
It seems when it is set to true it is not a problem, but when it isn't then some weirdness happens....

-Josh
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack/attachments/20120605/477b1222/attachment.html>