[openstack-dev] [Heat] DB schema changes for Convergence [was: Short term scaling strategies for large Heat stacks]

Clint Byrum clint at fewbar.com
Sat May 31 13:16:40 UTC 2014


Excerpts from Mike Spreitzer's message of 2014-05-31 07:28:16 +0100:
> Clint Byrum <clint at fewbar.com> wrote on 05/29/2014 09:09:18 PM:
> 
> > > > update-failure-recovery
> > > > =======================
> > > > 
> > > > This is a blueprint I believe Zane is working on to land in Juno. 
> ...
> > 
> > It's not just the observed state that you need in the database to 
> resume.
> > 
> > You also need the parameters and template snippet that has been
> > successfully applied.
> 
> Ah, right.  I revised the spec to add a table to hold the observed 
> parameters as individual records;
> we do not want to compose a whole "observed raw template" for each 
> observed state update.
> And you're right, the observed state needs more from the snippet; haven't 
> decided yet
> what I think about schema for the rest.

I think what we're saying is a bit skewed here.

We don't need an observed template or parameters for anything.
Observation is for things outside of Heat.

We do need to know what the currently active template snippet is for
the current update model as it must compute a diff and only apply diffs
based on the two templates. That is not observed though, that is simply
recorded and available inside Heat.

For convergence, we shouldn't need that. We will update the goal resource
snippet+params, and then use the observed state to know whether to create,
update, etc.



More information about the OpenStack-dev mailing list