Incremental backup of a volume failed using RBD.

[Zhao Liqiang]

When using rbd to backup a volume, the backup manager will attach the volume and make a temporary configuration file, then the driver execute the 'rbd export-diff' command using this configuration file. But I find that this file has been removed somewhere.

[Fikry (dxtxteam)]

Hi,I also met this problem .
https://bugs.launchpad.net/cinder/+bug/1680361

[tinytmy (tangmeiyan77)]

In the cinder
https://github.com/openstack/cinder/blob/master/cinder/backup/manager.py#L456-L459
function _attach_device use os-brick create and delete the temporary configuration file
https://github.com/openstack/os-brick/blob/master/os_brick/initiator/connectors/rbd.py#L108-L118

But after the temporary conf is used
https://github.com/openstack/cinder/blob/master/cinder/backup/drivers/ceph.py#L550-L589

The content of cmd1 is "['rbd', 'export-diff', '--id', 'cinder', '--conf', '/tmp/tmp9fsUYF', '--pool', 'volumes', '<email address hidden>', '-']"

when cmd1 and cmd2 executed, the ret is not none, raise a error, and set do_full_backup to true,
so alawys do full backup.

[Sean McGinnis (sean-mcginnis)]

Unassigning due to no activity for > 6 months.