[rhos-list] Running openshift on openstack

Matt Hicks mhicks at redhat.com
Thu Jan 17 14:19:04 UTC 2013


On 01/17/2013 09:05 AM, Steven Hardy wrote:
> Hi Matt,
>
> On Thu, Jan 17, 2013 at 08:38:38AM -0500, Matt Hicks wrote:
> <snip>
>> Steven and anyone else who is interested, could you send me your
>> GitHub id's so I can add you to the repository?  Right now the
>> script is in a private repository because it's with all the other
>> OpenShift Enterprise installation components but there is nothing
>> sensitive about the CloudFormations templates and we can certainly
>> pull it out.  I just figured the best way to look at it initially
>> would be with the kickstart scripts and some of the docs all
>> together and if you think it's helpful we can carve them out and
>> open source them in whatever way works best.
>>>> -Matt
> Thanks, my github id is "hardys" : https://github.com/hardys/
>
> I previously got access to your (now moved) enterprise-install repo, so I
> think I'm already a member of the openshift organization - so if you can
> just point me to the new location of the Cloudformation template that would
> be great.
Everything should be in here - 
https://github.com/openshift/enterprise/tree/enterprise-1.0/install-scripts/amazon

I noticed bleanhar has done the last few updates so he's probably the 
leading expert there now too.
>
> Last time I looked at the template, I noticed you're making use of
> AWS::Route53 resources, which we don't yet support in heat, so the first
> task for us will be to work out how to remove those and still
> install/configure things correctly for test purposes.
Yep, that will probably be the trickiest piece for us.  I'm still 
thinking about the best way to bring multiple machines up that all know 
about each other without fixed DNS addresses but I'm sure there is a 
way.  The main reason we use DNS is so that we can provisioning 
everything in 'parallel' (even though CloudFormations really doesn't) 
and when the machines come online, they already know how to connect to 
the other machines.  For example, when the broker is provisioned 
(broker.example.com), we configure it's connection to mongo by expecting 
the mongo machine to be at mongo.example.com. We'll just have a find a 
different way to do that.
>
> The other question I have is regarding the AMI you use - can you provide any
> details of this - e.g what version of RHEL (or ideally Fedora) do we need?
I did all of my testing with the latest RHEL and I essentially added 
cloud-init to it and burned my own AMI.  That was key to how we 
kickstarted the machines.  I believe Origin works on F17 right now but 
we have a known issue with F18.  Krishna - any updates there?
>
> Also in the openshift-amz.sh script which you wget into the template, is it
> OK to use the default repos_base, which appears to be a nightly build from a
> couple of months ago, or is there some other repo we should use instead?
We should eventually start using the coprs repository (3rd party 
repositories built through Koji) but it's not there yet.  In the 
meantime though, we might need to publish a temporary location since the 
yum repositories referenced in those are stale.  Krishna / Bill - any 
ideas here?
>
> Thanks for any info you can provide - will be great if we can get a better
> openshift-on-openstack (orchestrated by heat) demo capability sorted out!
>
> Steve

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/rhos-list/attachments/20130117/370f70bc/attachment.htm>


More information about the rhos-list mailing list