Comment 9 for bug 1818305

Agree with all this analysis, a head-to-head of a working v non working shows no difference [1]. One is dib 2.20.2 and other 2.20.3, only difference there is a change of a key for gentoo so it's not that. dib hasn't made any changes to this particular part of the code (block-device) for some time.

some thoughts

* -x logging should help see command output. dib could do better reporting in failure cases at the default level, however.
* working in /tmp ... slight chance that the loopback image that's made is too big for /tmp or similar? can set TMPDIR to a disk-based location
* possibly something hotplug-ish related that is noticing the new device, and doing ... something? to it. syslog might reveal something in this case?

[1] http://paste.openstack.org/show/747378/