How to reproduce:
1) Create one volume in one backend;
2) migrate this volume to SF backend;
3) attach this volume to any instance and SF driver would raise exception complaining something like:
"cinder-volume.log:37231:2014-10-15 23:36:11.205 59221 ERROR cinder.volume.drivers.solidfire [req-f6a2f0ef-2479-435f-a70e-2d56dc816925 ee0e38890e4942ef81614e194a0e253c 197048c1b08545e59ba7633ac9e17b65] Volume e8e35a66-5011-4179-b5b6-d893198b8fd6, not found on SF Cluster.
cinder-volume.log:37232:2014-10-15 23:36:11.206 59221 ERROR cinder.volume.drivers.solidfire [req-f6a2f0ef-2479-435f-a70e-2d56dc816925 ee0e38890e4942ef81614e194a0e253c 197048c1b08545e59ba7633ac9e17b65] Volume ID e8e35a66-5011-4179-b5b6-d893198b8fd6 was not found on the SolidFire Cluster!"
The reason migrated volume can't be attached to instances is because current Cinder migration generic workflow actually changes the UUID of volume, unfortunately SF relies on origin UUID of the volume to find it.
Fix proposed to branch: master /review. openstack. org/130313
Review: https:/