RE: Juju state as code?

2016-10-06 Thread Greg Mason
Patrik Karisch wrote:
> Hi there,
> 
> one thing which mocks my head, when evaluating Juju is, the state is only
> in the controller which gets deployed with commands. Which means I have
> IMHO no docs/history of how my current state of machines and charms was
> constructed?
> 
> As a side note: I'm used to Ansible and Puppet, which means I have all my
> infrastructe in a Git repository which describes my whole infrastructure
> and works always in a reproducable and idempotent way. How can I map this
> concept to Juju? I fear a little bit to loose this automation concept while
> using Juju.
> 
> Best regards,
> Patrik

Hi Patrik,

If you're looking for a repeatable way to deploy and test juju 
models/environments, you can also take a look at Mojo: 
https://mojo.canonical.com/

It allows you deploy specific versions of your code and charms, then run a 
specified set of tests after deployment. We use Mojo to deploy a large number 
of our juju environments. We test the deploy with Jenkins, and then (provided 
the CI deploy/test passes) deploy that commit/revision of the mojo spec to 
production. It's a good way to see how reproducible your environment is.

-Greg


--
Greg Mason
Cloud Reliability Engineer - Bootstack
Canonical, Ltd

-- 
Juju mailing list
Juju@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/juju


Re: Juju state as code?

2016-10-06 Thread Merlijn Sebrechts
Hi Patrik


Welcome to the Juju community, and thanks for the great question!

I see the bundle as a way to "save" the state of a model you constructed. A
bundle specifies a bunch of charms, their versions, their relations, and
their config values. You can find more info on bundles here:
https://jujucharms.com/docs/2.0/charms-bundles.

You can use the Juju GUI to create a bundle from the running model. You can
then redeploy that bundle whenever you need that model. Deploying bundles
can be done using command line tools or from the Juju GUI. The bundle is
just a yaml file so it's easy to put it in git and edit it yourself.

Does this answer your question?



Kind regards
Merlijn

2016-10-06 10:28 GMT+02:00 Patrik Karisch :

> Hi there,
>
> one thing which mocks my head, when evaluating Juju is, the state is only
> in the controller which gets deployed with commands. Which means I have
> IMHO no docs/history of how my current state of machines and charms was
> constructed?
>
> As a side note: I'm used to Ansible and Puppet, which means I have all my
> infrastructe in a Git repository which describes my whole infrastructure
> and works always in a reproducable and idempotent way. How can I map this
> concept to Juju? I fear a little bit to loose this automation concept while
> using Juju.
>
> Best regards,
> Patrik
>
> --
> Juju mailing list
> Juju@lists.ubuntu.com
> Modify settings or unsubscribe at: https://lists.ubuntu.com/
> mailman/listinfo/juju
>
>
-- 
Juju mailing list
Juju@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/juju


Juju state as code?

2016-10-06 Thread Patrik Karisch
Hi there,

one thing which mocks my head, when evaluating Juju is, the state is only
in the controller which gets deployed with commands. Which means I have
IMHO no docs/history of how my current state of machines and charms was
constructed?

As a side note: I'm used to Ansible and Puppet, which means I have all my
infrastructe in a Git repository which describes my whole infrastructure
and works always in a reproducable and idempotent way. How can I map this
concept to Juju? I fear a little bit to loose this automation concept while
using Juju.

Best regards,
Patrik
-- 
Juju mailing list
Juju@lists.ubuntu.com
Modify settings or unsubscribe at: 
https://lists.ubuntu.com/mailman/listinfo/juju