Newbie Question: How do I replace a machine in a deployed Model?

[Dmitrii Shcherbakov]

Hi Raghurama,

> Does Juju controller monitor the cluster and request MaaS for a new
machine if it detects one of the machines is gone?

No, it doesn't.

> Even if this has to be done manually, I did not see a replace-machine
option to Juju.

There's no such functionality - either an operator needs to make a decision
to do it or you need an automated system to do that depending on some
custom logic.

> Only add and remove units and machines. How does this work?

Juju itself does not know anything about applications you deploy - any
application-specific knowledge must be present in charms.

What you are looking for is an orchestrator type of capability - it will be
a layer on top of Juju or a charm with 'super cow powers' (namely, with
admin access to a juju controller).

A proof of concept would be a charmscaler from elastisys - it talks to a
juju controller directly and scales based upon CPU usage from nodes
collected via telegraf:
https://github.com/elastisys/layer-charmscaler-base/blob/164d163b4104cc47dcb1a32019509ba3f61d91eb/config.yaml#L9-L28
https://github.com/elastisys/layer-charmscaler#how-the-charmscaler-operates
https://jujucharms.com/u/elastisys/

https://github.com/elastisys/bundle-autoscaled-kubernetes
https://jujucharms.com/u/elastisys/autoscaled-kubernetes/bundle/0

https://elastisys.com/cloud-platform-features/predictive-auto-scaling/

You could build your own orchestrator with help of
https://github.com/juju/python-libjuju depending on your criteria. The
whole system could look as follows:

telegraf with your own juju input plugin -> prometheus alerts ->
orchestrator -> juju controller

The telegraf plugin would query juju and/or MAAS periodically to determine
the number of non-failed workers and send those metrics to prometheus.

Googling a little bit, I have found somebody's http server
https://github.com/imgix/prometheus-am-executor that handles prometheus'
alerts sent as HTTP requests. That kind of HTTP server could well hold your
orchestration logic and use python-libjuju to perform the necessary
add-unit-on-failure operations.

To sum up: it is important to understand the difference between Juju and
charms. Juju itself doesn't know anything about application-specific logic
- charms do. Charm code is executed by Juju agents at certain events and
this is where application-specific logic is actually executed. Any
orchestration code must have admin access to the juju controller and
contain subjective logic about how to scale-up or scale-down your
application.

https://github.com/juju/juju/blob/develop/doc/architectural-overview.md#juju-components

I hope that helps.

Best Regards,
Dmitrii Shcherbakov

Field Software Engineer
IRC (freenode): Dmitrii-Sh

On Fri, Sep 1, 2017 at 7:34 PM, Raghurama Bhat <rbhat at proofpoint.com> wrote:

> Any comments?
>
>
>
> Thanks,
>
>
>
> --Raghu
>
>
>
> *From: *<juju-bounces at lists.ubuntu.com> on behalf of Raghurama Bhat <
> rbhat at proofpoint.com>
> *Date: *Thursday, August 31, 2017 at 8:07 AM
> *To: *"juju at lists.ubuntu.com" <juju at lists.ubuntu.com>
> *Subject: *Newbie Question: How do I replace a machine in a deployed
> Model?
>
>
>
> Hi,
>
>
>
> I have a newbie question. I deployed a two node Kubernetes Core Cluster
> using Juju  into a MaaS Setup.   Now if I one of the Machine has a hardware
> failure, What is the process for replacing it with another machine? Does
> Juju controller monitor the cluster and request MaaS for a new machine if
> it detects one of the machines is gone? Even if this has to be done
> manually, I did not see a replace-machine option to Juju. Only add and
> remove units and machines. How does this work?
>
>
>
> Thanks,
>
>
>
> --Raghu
>
> --
> Juju mailing list
> Juju at lists.ubuntu.com
> Modify settings or unsubscribe at: https://lists.ubuntu.com/
> mailman/listinfo/juju
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.ubuntu.com/archives/juju/attachments/20170902/14a709de/attachment.html>