use of ramdisk causes builds to fail on machines with 'only' 2GB RAM

[Robert Collins]

tmpfs defaults to allowing up to 50% RAM, a machine with 2GB ram gets a 1G ramdisk. a 1G image fails to build when we copy it to the loopback device, due to the losetup device being inside the ramdisk.

We could move the device outside of the ramdisk, or make the ramdisk optional.

We should definitely document in /README.md the presence and use of the ramdisk too, as well as the failure mode - which is the loopback device becoming readonly , with block errors in dmesg.

[Robert Collins (lifeless)]

Workarounds : http://paste.ubuntu.com/5624972/ or use a machine with more RAM.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/28105

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/28105
Committed: http://github.com/stackforge/diskimage-builder/commit/d9a2211d14913a0875c730261611214ca55bdedc
Submitter: Jenkins
Branch: master

commit d9a2211d14913a0875c730261611214ca55bdedc
Author: Clint Byrum <email address hidden>
Date: Thu May 2 14:02:38 2013 -0700

    Only use tmpfs if build machine has 4GB+ RAM.

    For machines with low RAM (or no access to /proc/meminfo) the builder
    will still run, but will just build inside the filesystem that is
    hosting /tmp. This will result in a slower build (especially if there
    are a lot of .deb packages installed).

    Fixes bug #1175453

    Change-Id: I79f2672058c11e377548820df0ab4fad8f47ffdc