When using whole disk images, no resize/expansion of any partitions, logical volumes, filesystems, etc. is performed. Therefore, the *virtual size* of the image will equal the amount of disk that one wants to use on one's overcloud nodes. For example, one might end up with a overcloud-full-partitioned.qcow2 file like this: image: overcloud-full-partitioned.qcow2 file format: qcow2 virtual size: 100G (107374182400 bytes) disk size: 7.0G cluster_size: 65536 Format specific information: compat: 1.1 lazy refcounts: false refcount bits: 16 corrupt: false This files is 7.0G, but Ironic will attempt to convert this image as a 100G raw image when it is first used to provision an overcloud node. This can easily lead to a deployment failure with a bunch of 'No valid host was found' errors. The actual error (in nova-conductor.log) is: Error: Disk volume where '/var/lib/ironic/master_images/tmpoXvZr2' is loca ted doesn't have enough disk space. Required 102400 MiB, only 61590 MiB available space present. I.e. Ironic needs 100G of space under /var/lib/ironic/master_images on the undercloud.
Upon further investigation, I have determined that the full logical size of the image is not really required. When Ironic calls qemu-img to convert the image from QCOW to raw, it creates a *sparse* file, which uses about the same amount of actual disk space as the QCOW image. So this looks like a bug-bug, not just a doc bug.
(In reply to Ian Pilcher from comment #1) > So this looks like a bug-bug, not just a doc bug. Filed here - https://bugzilla.redhat.com/show_bug.cgi?id=1535637
Scoping old bugs. Closing this one because: a) It looks like it's being handled on the engineering side b) We added a note regarding the raw size of the disk that says: == Important == When the director deploys the overcloud, it creates a RAW version of the overcloud image. This means your undercloud must have necessary free space to accommodate the RAW image. For example, if you increase the security hardened image size to 40G, you must have 40G of space available on the undercloud’s hard disk. https://access.redhat.com/documentation/en-us/red_hat_openstack_platform/12/html/director_installation_and_usage/appe-whole_disk_images