Open Stack

On 23/02/16 15:58, Ludwig Tirazona wrote:
> Hello Everyone,
>
> 	I want to understand the Architecture a bit more. Thanks in advance for
> any answers.
>
> 	1.) The docs tell me that trove-guestagent connects to a RabbitMQ host
> to listen for instructions. Is this RabbitMQ host typically the one that
> the other OpenStack services connect to?
>
> 	2.) If yes, how does that normally occur, given that the Trove Instance
> is usually isolated in its own tenant network, and my RabbitMQ is
> located on the Internal Network (no public access)?
>
> 	3.) I'm trying to make my own Guest Image, and have found the
> documentation to be lacking. I've momentarily given up on making one
> using the diskimage-builder and trove-integration. Is it "acceptable" to
> just use an Ubuntu Cloud Image, install MySQL and trove-guestagent on
> it, then use that as a Datastore? Is that a sane route to take, or am I
> missing something important?
>
>

Good questions!

1/ You can (and should) use a different rabbit (or at least a different 
user if the same rabbit)

2/ Umm, don't make it the same in general...and yes the docs are 
strangely silent here! There is some work underway to clarify and 
improve the security separation.

3/ Yes, this is fiddly - need to make sure that user names etc match, 
otherwise spawned instance just sits in 'BUILDING' forever (I think I've 
posted on this issue during the Juno cycle...on this list too I think - 
have a search for more details). I managed to get mine working using 
disk-image-builder but things could have changed.

Overall, the setup is quite complex and unforgiving, and you have 
clearly identified the murky areas that could benefit from some doc 
revision!

regards

Mark