ARCHIVED / NO LONGER RELEVANT!

NOTE

This is just a spike of some templating and binding, definitely not sold on the formats.

Declarative Continuous Delivery in Spinnaker

example configuration for deployment targets and pipeline templates and pipeline bindings

entrypoint

PipelineBindingSpec loads a pipeline binding for gate, asks for the "Deploy to prestaging" pipeline, and renders the resulting execution json.

implementation

HierarchicalContent supports generic layerable maps with parameter replacement (used for deployment target)
PipelineTemplate is the definition of a pipeline and the expected bindings
- exposes a method to produce a pipeline execution, given the concrete deployment targets
PipelineBindings merges the deployment targets and pipeline templates

deployment targets:

accounts.json defines account specific configurations to mix in (AZs per region), keypair, cloudprovider
deploydefaults.json defines a generic deploy stage configuration
concrete implementations in gate.json and clouddriver.json mix in deploydefaults.json and accounts.json to produce concrete targets for prestaging and main stacks

pipeline templates:

spinnakerPipelines.json defines a generic bake + tag, promote to prestaging, promote to main flow

pipeline bindings

gatePipelineBindings.json brings in gate.json and spinnakerPipelines.json and maps the appropriate gate deployment targets for the standard spinnaker flow

(NOT IMPLEMENTED) clouddriverPipelineBindings.json brings in clouddriver.json and spinnakerPipelines.json but for the deploy to main stage, extends the pipeline to support the redis flipping case (prestaging just does a standard deploy to stack 'a')

functions / property placeholders in json

property placeholders

deployment target spike uses property placeholder approach (mustache style {{ }}) to inject property values

(NOT IMPLEMENTED) functions as object attributes:

as an attempt to preserve valid json syntax, these functions are defined as a function attribute that operates on the object

#filter(key, subKey): operates on key within this object, retaining only subKey (See accounts.json, filters az map by region)

#foreach(deploymentTarget, newName): - for each cluster in the deployment target deploymentTarget, emit a copy of the object, with a new deployment target named newName available for templating

functions on deployment targets:

pipeline templates uses approach of defining functions that operate on the deployment targets

//TODO - some of these seem like they expect single cluster vs multiple, spike implementation just throws an exception if multiple distinct values would result

#account - the account name for the deployment target
#stack - the stack for the deployment target
#application - the application 
#regions - list of regions 
#clusters - list of cluster definitions
#clusterName - name of the cluster
#cloudProviderType - cloud provider of the cluster

spinnaker/dcdspike