[openstack-dev] [Mistral] Crack at a "Real life" workflow

Renat Akhmerov rakhmerov at mirantis.com
Thu Mar 6 10:27:10 UTC 2014

> How is this different then chef/puppet/ansible …?
> Forgive me if that has already been answered, I didn't see an answer to that under the FAQ at https://wiki.openstack.org/wiki/Mistral
No problem. Actually forgive us if information on wiki is not enough to understand the purpose of the technology. We’re now fully in development and wiki doesn’t match the current state of the code on 100%.

Mistral is a general purpose workflow engine, not a deployment tool like chef or puppet. It’s useful when:
We need to implement a long-lived process (what we call a workflow) spanning multiple distributed components (in our case VMs and services in OpenStack).
This process can be decomposed into multiple individual tasks.
We need to have full control over the life-cyle of this process (stop/resume/destroy).
We need to be able to track the state of the process (is it in failure, finished successfully, still in progress and so on) and history of previous running processes.
We need to be able to rollback the process (execute in reverse order).
We want to run the process on schedule or external event.
We want to use any language to interact with the process (not only Python).

So a deployment tool could use Mistral to implement its logic internal logic: represent a deployment process as a workflow and have Mistral take care of its execution.

I think it would also be useful to take a look at this presentation: http://www.slideshare.net/RenatAkhmerov/mistral-openstack-meetup-feb-5

Please also refer to the description of Mistral use case on wiki.

If you need more information or you have some specific questions feel free to ask.


Renat Akhmerov
@ Mirantis Inc.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20140306/019c137d/attachment.html>

More information about the OpenStack-dev mailing list