Bug 1466391

Summary: Image replication and synchronization across multiple Ceph Backend Stores.
Product: Red Hat OpenStack Reporter: Bertrand <brault>
Component: openstack-glanceAssignee: Cyril Roelandt <cyril>
Status: CLOSED DUPLICATE QA Contact: Mike Abrams <mabrams>
Severity: high Docs Contact:
Priority: high    
Version: 17.0 (Wallaby)CC: abdelhadi.chari, akekane, cschwede, dasmith, ealcaniz, ecv-redhat, eglynn, ekuvaja, gcharot, jcoufal, johfulto, mabrams, owalsh, scohen, shan, srevivo, tshefi
Target Milestone: ---Keywords: FutureFeature, Triaged
Target Release: ---   
Hardware: Unspecified   
OS: Unspecified   
Whiteboard:
Fixed In Version: Doc Type: If docs needed, set a value
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2019-10-04 05:40:30 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:
Bug Depends On: 1225775, 1466008, 1598716, 1712001    
Bug Blocks: 1459148, 1476902, 1496482, 1707841, 1897605    

Description Bertrand 2017-06-29 13:59:16 UTC 
In the distributed compute use case, the single control zone hosting the controllers is separated from the multiple compute zones hosting the compute nodes.
It is assumed here that Glance will use multiple Ceph Backend Stores in respective compute zone to minimize image transfer and thus decrease instance spin up time.
In such situation it is important to be able to upload images once and replicate those on multiple Glance Backend stores as well as keeping those in sync on all remote compute zones.

Thus VM images made available through Glance will need to be replicated and synchronized on the multiple respective Ceph Backend Stores on respective compute zones.

One of the reason is that downloading an image from the central control zone over a WAN to a remote compute zone would be time consuming. 
It is critical to be able to spin up new VMs (VNFs) quickly as it is expected to launch new instances in the compute zone in less that a minute or so. 

This RFE depends on Bug 1225775 - [RFE] Glance should be able to manage multiple Ceph clusters as backends


Similar blueprint:
https://blueprints.launchpad.net/glance/+spec/image-bits-replication

Also in term of image replication, there is: glance-replicator utility.
https://docs.openstack.org/developer/glance/cli/glancereplicator.html
Comment 1 Sean Cohen 2017-08-09 13:21:29 UTC 
Pushing out to OSP14 for review,

I suggest to investigate Images synchronization at Kingbird level 

[1] https://blueprints.launchpad.net/kingbird/+spec/image-synchronization
[2] https://review.openstack.org/#/c/462551/

Sean
Comment 4 Bertrand 2018-05-30 17:04:35 UTC 
Also referencing a discussion that took place during Openstack Summit:
https://etherpad.openstack.org/p/yvr-edge-cloud-images
Comment 5 Tzach Shefi 2018-07-18 14:28:01 UTC 
I'm thinking replication settings call for a new/missing THT RFE support, right? 

If you share my feeling, open a new THT RFE and set depends on flag. 
As the multi Ceph THT (bz1466008) I just added won't configure replication. 
Or maybe the replication THT will do it all.  
I don't yet know but want to get THTs started.

"Done" means done via OPSD.
Comment 6 Sean Cohen 2018-08-13 13:16:04 UTC 
The goal of this bug is to track the testing of Ceph image replication in a multiple Ceph Clusters as backends, per bug 1225775
Comment 16 Bertrand 2019-10-01 07:34:12 UTC 
Hi Erno,
Are we getting this implemented in OSPS17?
Thx,
Bertrand
Comment 17 Abhishek Kekane 2019-10-04 05:40:30 UTC 
This BZ is marked duplicate of #1758416 and will be closed. The current BZ talks about the replication of image across multiple ceph stores only where as the new BZ #1758416 will able to replicate image across different kind of backends.

*** This bug has been marked as a duplicate of bug 1758416 ***
Comment 18 Red Hat Bugzilla 2023-09-18 00:12:32 UTC 
The needinfo request[s] on this closed bug have been removed as they have been unresolved for 120 days