[Openstack] compiler for heat templates

Zane Bitter zbitter at redhat.com
Tue Mar 20 14:56:26 UTC 2018


On 20/03/18 04:53, Sashan Govender wrote:
> Hi
> 
> Is there a way to check heat templates. At the moment I run one and it 
> errors at runtime when, for example an something expects a string but 
> gets a list. For example in this case of an OS::Heat::SoftwareConfig 
> resource, the config attribute below expects a string, which is why 
> str_replace works

Note that str_replace doesn't have to return a string (it returns 
whatever type its 'template' is), but yeah in this case it does so it works.

>   some_resource:
>        type: OS::Heat::SoftwareConfig
>        properties:
>          config:
>            str_replace:
>              params:
>                $repstr$:
>                  list_join: ['-', [ {get_param: cluster_name}, 'xyz']]
>              template: |
>                #!/bin/bash
>                echo $repstr$ >> /etc/somefile
> 
> According to this 
> https://docs.openstack.org/heat/latest/template_guide/openstack.html#OS::Heat::SoftwareConfig
> 
> the config property expects a string.

Yep. OS::Heat::StructuredConfig is what you want if you want to pass 
lists or maps.

> If I replace str_replace with something that generates a list (e.g. 
> repeat) it fails at runtime. Is there a way to type check this? I tried 
> 'heat template-validate' but it didn't do what I expected...

template-validate unfortunately doesn't validate property _values_ at 
all, for historical reasons (in CloudFormation the parameter values 
don't get passed to template-validate, although in the OpenStack API 
they optionally can be since Icehouse). I think this is something we 
should look at improving. I filed a bug report:

https://bugs.launchpad.net/heat/+bug/1757163

An additional subtlety is that because the output of intrinsic functions 
can't be known ahead of time in many cases (e.g. get_resource, 
get_attr), we still have to defer some checking to resource 
create/update time when all of the data is available. We do have some 
thoughts around how to improve this in the future, but it's a fairly 
major development effort.

cheers,
Zane.



More information about the Openstack mailing list