Bug 1658833

Summary:	sparseness is not preserved across online volume migration - CEPH
Product:	Red Hat OpenStack	Reporter:	Siggy Sigwald <ssigwald>
Component:	openstack-cinder	Assignee:	Jon Bernard <jobernar>
Status:	CLOSED NOTABUG	QA Contact:	Tzach Shefi <tshefi>
Severity:	high	Docs Contact:	Kim Nylander <knylande>
Priority:	high
Version:	10.0 (Newton)	CC:	jdurgin, johfulto, lhh, mabrams, nlevine, tenobreg, tshefi
Target Milestone:	---	Flags:	tshefi: automate_bug-
Target Release:	---
Hardware:	x86_64
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	If docs needed, set a value
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2019-01-09 16:47:01 UTC	Type:	Bug
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:

Description Siggy Sigwald 2018-12-12 23:36:42 UTC

Description of problem:
Volume migration converts provision to thick causing OSD full
I executed for i in `rbd ls volumes-fast`;do rbd du volumes-fast/$i;done. As you see below 5 (100 GB ) + 5 (300GB) = 2000GB. The 100 Gb volumes are volumes migrated from normal ceph volume type to fast volume type. Now what I see is that when migration of volumes across volume type happens CEPH is treating them as thick provision and provisioned = used

Comment 3 Jon Bernard 2018-12-19 15:17:57 UTC

It's likely that the volume migration is not optimized, and a full copy is expected in the current state.  With the cinder configuration and log files included I could tell you for sure.

This is a duplicate of https://bugzilla.redhat.com/show_bug.cgi?id=1655490

Comment 5 Telles Nobrega 2019-01-09 16:47:01 UTC

Hi, After discussion with Jon Bernard, we decided that this is not a bug, the behavior is as expected but it could be improved, so I'm closing this as not a bug and it can be reopened later as an RFE.

Comment 6 Tzach Shefi 2019-01-29 07:11:00 UTC

Close loop wise as behavior is as expected -> NOTABUG. 
Nothing for QE to test/automate.