Open Stack

The training-labs scripts currently install OpenStack with this config:

controller: 1 processor, 1 GB RAM
network: 1 processor, 512 MB RAM
compute: 2 processors, 1 GB RAM

Main issue is that you can't start a whole lot of instances on compute with
1 GB RAM. Then again, you catch some interesting corner cases.

So yes, this does work for a proof of concept environment. I suspect a)
that having 2 processors on compute is not strictly necessary and b) it
will still work once we're done adding heat and ceilometer (in progress).

I wouldn't recommend putting any serious load on such an environment,
though, unless you are studying failure modes.

Roger

On Wed, 15 Apr 2015 11:22:37 +0200, Christian Berendt wrote:
> At the moment we mention the following minimum requirements for a
> proof-of-concept environment in chapter 2 (basic environment):
> 
> ---snip---
> Controller Node: 1 processor, 2 GB memory, and 5 GB storage
> 
> Network Node: 1 processor, 512 MB memory, and 5 GB storage
> 
> Compute Node: 1 processor, 2 GB memory, and 10 GB storage
> ---snap---
> 
> Does this really work? Has anybody tried to run a network node with only
> 512 MByte memory? Or a controller node with RabbitMQ, MySQL, MongoDB and
> a lot of OpenStack services with 1 VCPU and 2 GByte memory?
> 
> I would propose to increase the minimum requirements:
> 
> Controller Node: 4 processors, 8 GB memory, and 15 GB storage
> Network Node: 2 processors, 4 MB memory, and 10 GB storage
> Compute Node: 2 processors, 4 GB memory, and 20 GB storage
> 
> Christian.