I poked at this with some additional volumes attached to the cloud instance:
Offending device: ubuntu@juju-t-machine-22:~$ sudo umount /dev/vdb ubuntu@juju-t-machine-22:~$ sudo lsof | grep vdb jbd2/vdb- 1268 root cwd DIR 253,1 4096 2 / jbd2/vdb- 1268 root rtd DIR 253,1 4096 2 / jbd2/vdb- 1268 root txt unknown /proc/1268/exe Additional device: ubuntu@juju-t-machine-22:~$ sudo mount /dev/vdc /mnt2 ubuntu@juju-t-machine-22:~$ sudo lsof | grep vdc jbd2/vdc- 16058 root cwd DIR 253,1 4096 2 / jbd2/vdc- 16058 root rtd DIR 253,1 4096 2 / jbd2/vdc- 16058 root txt unknown /proc/16058/exe ubuntu@juju-t-machine-22:~$ sudo umount /dev/vdc ubuntu@juju-t-machine-22:~$ sudo lsof | grep vdc As you can see, the jbd2 process for vdb appears to hang around, which I think is what is keeping the partition table locked in kernel and hence stale. ** Also affects: linux (Ubuntu) Importance: Undecided Status: New -- You received this bug notification because you are a member of Ubuntu Server Team, which is subscribed to ceph in Ubuntu. https://bugs.launchpad.net/bugs/1371526 Title: ceph-disk-prepare command always fails; new partition table not avaliable until reboot To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/ceph/+bug/1371526/+subscriptions -- Ubuntu-server-bugs mailing list Ubuntu-server-bugs@lists.ubuntu.com Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/ubuntu-server-bugs