Migration from essex to havana

asked 2013-10-31 10:19:21 -0600

updated 2014-01-22 23:54:55 -0600

smaffulli gravatar image


I know that this question about migration from folsom to grizzly remains unanswered but I'll give it a go anyway.

We have an essex infrastructure in place. Any chance of migrating bits of this to a newer version of OpenStack ? If not, are there any recommended guidelines on how to make two infrastructures co-exist and the steps to take to "manually" migrate some machines from one to the other ?

edit retag flag offensive close merge delete

2 answers

Sort by ยป oldest newest most voted

answered 2013-11-01 11:20:21 -0600

larsks gravatar image

updated 2014-01-22 23:55:23 -0600

smaffulli gravatar image

I don't think you can migrate "bits of it". I think you would probably need to upgrade the whole thing at once.

Setting up a parallel environment in a more recent version is probably the easiest way forward. Then you can snapshot instances in your essex environment, manually send the snapshots to glance in your new environment, and then boot them there. Once you've moved everything over, you can tear down the essex environment and add the equipment to your new deployment.

It's possible that you could upgrade essex -> folsom -> grizzly -> havana, but I have no idea if that's a sane path or not. You'd want to be very sure about your backups.

edit flag offensive delete link more

answered 2013-11-01 15:00:27 -0600

updated 2014-01-22 23:55:36 -0600

smaffulli gravatar image


The best approach I'm aware of is to setup a new havana-based cloud right beside the old one (pull nodes off the old one, if needed, to give the new one enough resources) - and then migrate your application workloads over to the new cloud.

This way, you're not really caring about the cloud instances themselves. They were just workers - so you replace them with new workers in the new cloud, doing the same job for your end users that the old workers were doing.

With puppet and friends it's possible to deploy production nodes "hands off" - i.e. no human ever logs in on them. Very possibly you could use the same puppet deployment scripts with Havana that you may have originally targetted to Essex.

Also, LBaaS can help minimize the impact of switchover. Hope this helps!

edit flag offensive delete link more

Get to know Ask OpenStack

Resources for moderators

Question Tools



Asked: 2013-10-31 10:19:21 -0600

Seen: 426 times

Last updated: Jan 22 '14