[Openstack-operators] Does anyone copy images between multiple glance instances?

Abel Lopez alopgeek at gmail.com
Tue May 20 21:59:07 UTC 2014


Firstly, make sure you’re using the latest glance client ( https://github.com/openstack/python-glanceclient )
Also, make sure you’re an admin user

On May 20, 2014, at 2:55 PM, Scott Devoid <devoid at anl.gov> wrote:

> Abel, Warren,
> 
> Thanks for the quick replies! We are ok with running a script to copy images from one glance to another. Eventual consistency is ok and we only need this to run until we can shutdown the Essex era glance. But the problem I have is that the glance-client doesn't give me full visibility into an image's attributes, or the API won't let me set attributes (like snapshot instance_uuid).
> 
> I am hoping that I'm being dumb or this is a lack of documentation issue.
> 
> ~ Scott
> 
> 
> On Tue, May 20, 2014 at 4:49 PM, Warren Wang <warren at wangspeed.com> wrote:
> I think what you ultimately want is a unified glance infrastructure, and we're in the same boat. Currently, I don't know of a good way to do this without taxing the network links, or slowing down the snapshots drastically. For now, we have settled on manually syncing the base instances from one region to the others.
> 
> Warren
> 
> Warren
> 
> 
> On Tue, May 20, 2014 at 5:41 PM, Scott Devoid <devoid at anl.gov> wrote:
> We have two glance instances in separate deployments (an essex deployment and a havana deployment) and we would like to synchronize images between the two.
> 
> Uploading the raw image data works fine with either the command line or the API, but making the metadata consistent is quite difficult. In the essex API I can't even find container_format or disk_format reported in the instance.
> 
> I also can't find a way with the client to affix the snapshot metadata on the Havana side other than through a direct HTTP PATCH request, and it is not very clear what headers that request is expecting. [1]
> 
> Another thought I had, since both instances of glance are using the Filesystem store (w/ separate NFS deployments) was to simply use the existing NFS in essex and inform the havana glance about the existence of an image through the --location tag. However this is intentionally blocked. [2]
> 
> Thank you in advance for the help,
> ~ Scott
> 
> [1] http://developer.openstack.org/api-ref-identity-v2.html
> [2] https://bugs.launchpad.net/glance/+bug/942118
> 
> 
> 
> _______________________________________________
> OpenStack-operators mailing list
> OpenStack-operators at lists.openstack.org
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators
> 
> 
> 
> _______________________________________________
> OpenStack-operators mailing list
> OpenStack-operators at lists.openstack.org
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-operators/attachments/20140520/edfee378/attachment.html>


More information about the OpenStack-operators mailing list