Comment 6 for bug 1394179

[Oops, my previous comment got added twice.]

Jay, you're right: I indeed reboot the DevStack instance, and did ./rejoin-stack.sh. I just did it again:

 - Reboot the VM
 - SELinux in Permissive mode
 - Rejoin the Stack service: $ ./rejoin-stack.sh

   Fails with the same stack trace (looking at the c-api `screen` window) as noted in Description.

And yep, seems like it lost the mounted loop device:

  $ losetup -a
  $

When I'm not using DevStack, I create loop device by hand, something like:

  $ dd if=/dev/zero of=/cinder-volumes bs=1 count=0 seek=5G
  $ losetup -fv /cinder-volumes
  $ losetup -l
  $ vgcreate cinder-volumes /dev/loop0
  $ vgdisplay cinder-volumes

Probably I should do something like that in this case.

Jay, feel free to close this if you think DevStack shouldn't do it, when you rejoin-stack.sh. Thanks for helping to debug this.