Cinder

Adjust ceph.backup driver behavior if inremental backup fails

Registered by Vladimir Prokofev

When using cinder-backup with CEPH driver in some situations it can fail to create thin backup with "aborting incremental backup" log message.
As example, if you have a 500GB volume with only 10GB used, full thin backup would take only 10GB storage in CEPH pool, while full thick backup will take 500GB. This will take more time, and potentially can lead to CEPH cluster DoS due to "osd full" situation because of planned storage space calculation misjudgement.
I propose to add a boolean flag "--full-if-incremental-fail" to manage behavior in this situation - "true" to create a full thick backup, and "false" to abort operation, defaulting to "true".

Blueprint information

Status:
Complete
Approver:
None
Priority:
Undefined
Drafter:
Vladimir Prokofev
Direction:
Needs approval
Assignee:
None
Definition:
Obsolete
Series goal:
None
Implementation:
Unknown
Milestone target:
None
Completed by
Sean McGinnis

Related branches

Sprints

Whiteboard

We don't want to change current behavior or requiring a full backup before an incremental can be performed.

(?)

Work Items

This blueprint contains Public information 
Everyone can see this information.
Edit subscription

Subscribers

No subscribers.