[Openstack] Making Image Launch faster in Openstack

Gary Kotton gkotton at vmware.com
Mon May 26 13:54:20 UTC 2014

At the summit there was a session about this. Please also see https://review.openstack.org/85792

From: Rakesh Sinha <rakesh.sinha at gslab.com<mailto:rakesh.sinha at gslab.com>>
Date: Monday, May 26, 2014 4:35 PM
To: Pádraig Brady <P at draigbrady.com<mailto:P at draigbrady.com>>
Cc: "openstack at lists.openstack.org<mailto:openstack at lists.openstack.org>" <openstack at lists.openstack.org<mailto:openstack at lists.openstack.org>>
Subject: Re: [Openstack] Making Image Launch faster in Openstack

Thanks Padraig, but is there any other way than using "squid". Like calling API's in glance or calling nova-compute API's or add something in glance code which triggers copy of vmdk to all compute nodes after image is uploaded.

On Mon, May 26, 2014 at 5:47 PM, Pádraig Brady <P at draigbrady.com<mailto:P at draigbrady.com>> wrote:
On 05/26/2014 12:54 PM, Rakesh Sinha wrote:
> Hi
> I am aware that first instance launch over a Compute Node takes time because it has to copy image from the Glance Server to the Compute Node.
> Since the local copy is available over the Compute now so next time it launches faster.
> I want that when the Image is uploaded over the Glance Server then only it makes a copy of it over the Compute Server also so that there is no delay even if the Image is launched for the first time or nth time.
> Please suggest proper methods to achieve so or maybe some code changes.

CERN did some work on that: http://coarasa.dyndns.org/?p=162<https://urldefense.proofpoint.com/v1/url?u=http://coarasa.dyndns.org/?p%3D162&k=oIvRg1%2BdGAgOoM1BIlLLqw%3D%3D%0A&r=eH0pxTUZo8NPZyF6hgoMQu%2BfDtysg45MkPhCZFxPEq8%3D%0A&m=uk5UeNLCDtH4%2Bx1kXlhPNZ8GLyRTbsDUSKlzpqsezDY%3D%0A&s=b76ce857ca16d18b390c9fd45d1f6282dbf0a49dfbc79bfdea724b1f94837e09>


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack/attachments/20140526/22d6cf30/attachment.html>

More information about the Openstack mailing list