Parameters, Credentials, Outputs, and Images in Porter
In the Porter manifest, you can declare both parameters and credentials. In addition to providing a mechanism for declaring parameters and credentials at the bundle level, Porter provides a way to declare how each of these are provided to mixins. This mechanism is also applicable to declaring how output from one mixin can be passed to another, as well as how to consume parameters, credentials and outputs from bundle dependencies. Finally, you can also use this technique to reference images defined in the images
section of the manifest.
- Wiring Installation Metadata
- Parameters
- Credentials
- Outputs
- Wiring Custom Metadata
- Wiring Images
- Wiring Dependency Outputs
- Combining References
Wiring Installation Metadata
The installation name is available at runtime as {{ installation.name }}
. In the example below, we install a helm chart
and set the release name to the installation name of the bundle:
install:
helm:
description: Install myapp
name: "{{ installation.name }}"
chart: charts/myapp
Parameters
In order to declare a parameter in a Porter bundle, you first declare a parameters block with one or more parameters in the porter.yaml
. For example, to declare a parameter named database_name, you might include the following YAML block:
parameters:
- name: database_name
type: string
This is the minimum required to create a parameter in Porter. Porter will specify an environment variable destination that defaults to the upper-cased name of the parameter.
You can also provide any other attributes, as specified by the CNAB parameters specification. To specify a default value, for example, you could provide the following parameter definition:
- name: database_name
type: string
default: "wordpress"
File Parameters
Porter also enables the use of file parameters in a bundle.
For instance, a bundle might declare a parameter mytar
of type file
, to exist at /root/mytar
in the execution environment:
- name: mytar
type: file
path: /root/mytar
which can be used in a step like install
:
install:
- exec:
description: "Install"
command: bash
flags:
c: tar zxvf /root/mytar
The syntax to pass a parameter to porter is the same for both regular and file parameters:
$ porter install --param mytar=./my.tar.gz
See the Parameters section of the Author Bundles doc for additional examples and configuration.
Wiring Parameters
Once a parameter has been declared in the porter.yaml
, Porter provides a simple mechanism for specifying how the parameter value should be wired into the mixin. To reference a parameter in the bundle, you use a notation of the form "{{ bundle.parameters.PARAM_NAME }}"
. This can be used anywhere within step definition. For example, to use the database_name
parameter defined above in the set
block of the helm mixin:
install:
- helm:
description: "Install MySQL"
name: porter-ci-mysql
chart: stable/mysql
version: 0.10.2
replace: true
set:
mysqlDatabase: "{{ bundle.parameters.database-name }}"
mysqlUser: "root"
Or to provide a parameter to the command
attribute of the exec mixin:
parameters:
- name: command
type: string
default: "echo Hello World"
install:
- description: "Install Hello World"
exec:
command: "{{ bundle.parameters.command }}"
This syntax is used in dictionary, as above, or in a list:
parameters:
- name: command
type: string
default: "echo Hello World"
install:
- description: "Install Hello World"
exec:
command: bash
flags:
c: "{{ bundle.parameters.command }}"
NOTE: These references must be quoted, as in the examples above.
See Parameters to learn how parameters are passed in to Porter prior to bundle execution.
Credentials
Credentials are defined in the porter.yaml
with a YAML block of one more credential definitions. You can declare that a credential should be placed in a path within the invocation image or into an environment variable.
To declare a file injection:
credentials:
- name: kubeconfig
path: /root/.kube/config
To declare an environment variable injection:
credentials:
- name: SUBSCRIPTION_ID
env: AZURE_SUBSCRIPTION_ID
Wiring Credentials
The same mechanism for declaring how to use a parameter can be used for credentials. To declare a credential usage, references are defined with the following syntax: "{{ bundle.credentials.CREDENTIAL_NAME}}"
.
When the bundle is executed, the Porter runtime will locate the parameter definition in the porter.yaml
to determine where the parameter value has been stored. The Porter runtime will then rewrite the YAML block before it is passed to the mixin. See Credentials to learn how credentials work.
Outputs
In addition to parameters and credentials, Porter introduces a type called an output. Outputs are values that are generated during the execution of a mixin. These could be things like a hostname for a newly provisioned Azure service or a generated password from something installed with Helm. These often need to be used in subsequent steps so Porter allows you to declare how to reference them after they have been created. The form used to declare these varies by mixin, but might look something like:
install:
- arm:
description: "Create Azure MySQL"
type: arm
template: "arm/mysql.json"
name: demo-mysql-azure-porter-demo-wordpress
resourceGroup: "porter-test"
parameters:
administratorLogin: "{{ bundle.parameters.mysql_user}}"
administratorLoginPassword: "{{ bundle.parameters.mysql_password }}"
location: "eastus"
serverName: "mysql-jeremy-porter-test-jan-2018"
version: "5.7"
sslEnforcement: "Disabled"
databaseName: "{{ bundle.parameters.database_name }}"
outputs:
- name: "MYSQL_URL"
key: "MYSQL_HOST"
In this example, a new output will be created named MYSQL_URL
. The Azure mixin allows you to specify the key to fetch the output from, in this case it is MYSQL_HOST
. Each mixin can provide different ways of addressing outputs, so refer to the schema for each mixin. The Porter runtime will keep a map in memory with each of the outputs declared.
Wiring Outputs
Once an output has been declared, it can be referenced in the same way as parameters and credentials. Outputs are referenced with the syntax "{{ bundle.outputs.OUTPUT_NAME }}"
For example, given the install step above, we can use the MYSQL_URL
with the helm mixin in the following way:
- helm:
description: "Helm Install Wordpress"
name: porter-ci-wordpress
chart: stable/wordpress
set:
mariadb.enabled: "false"
externalDatabase.port: 3306
externalDatabase.host: "{{ bundle.outputs.MYSQL_URL }}"
externalDatabase.user: "{{ bundle.parameters.mysql_user }}"
externalDatabase.password: "{{ bundle.parameters.mysql_password }}"
externalDatabase.database: "{{ bundle.parameters.database_name }}"
Just like in the case of credentials and parameters, the value of the bundle.outputs.MYSQL_URL
reference will be rewritten in the YAML before the helm mixin is invoked.
Parameters can also use the value from an output from the current bundle or one of its dependencies as its default value
using the source
field when defining the parameter.
Source an output from the current bundle
parameters:
- name: tfstate
type: file
path: /cnab/app/tfstate
source:
output: tfstate
Source an output from a dependency
parameters:
- name: connection-string
type: string
source:
dependency: mysql
source: connstr
Wiring Custom Metadata
In the porter.yaml
, you can define custom metadata and use it in your bundle.
These custom values are hard-coded into the bundle and cannot be modified at
runtime. If you need that, then you should use a parameter.
custom:
myApp:
featureFlags:
featureA: true
Now you can use the custom values in your actions like so:
install:
helm:
description: Install myapp
chart: charts/myapp
set:
featureA: "{{ bundle.custom.myApp.featureFlags.featureA }}"
Wiring Images
In the porter.yaml
, you can define what images will be used within the bundle with the images
section:
images:
ALIAS:
description: A very useful image
imageType: docker
repository: gcr.io/mcguffin-co/mcguffin
digest: sha256:85b1a9
tag: v1.1.0
These images will be used to build the bundle.json
images section, but can also be referenced using the same syntax you would use for referencing parameters
, credentials
, and outputs
.
- helm:
description: "Helm Install Wordpress"
name: porter-ci-wordpress
chart: stable/wordpress
set:
image.repository: "{{ bundle.images.ALIAS.repository }}"
image.tag: "{{ bundle.images.ALIAS.tag }}"
Wiring Dependency Outputs
You can reference outputs from a dependency defined in your bundle using the syntax {{ bundle.dependencies.DEPENDENCY.outputs.OUTPUT }}
.
For example, consider a bundle that creates a mysql defined with the following porter.yaml
:
name: mysql
version: 0.1.3
tag: getporter/mysql
mixins:
- helm
credentials:
- name: kubeconfig
path: /root/.kube/config
parameters:
- name: database-name
type: string
default: mydb
env: DATABASE_NAME
- name: mysql-user
type: string
env: MYSQL_USER
install:
- helm:
description: "Install MySQL"
name: porter-ci-mysql
chart: stable/mysql
version: 1.6.2
replace: true
set:
mysqlDatabase: "{{ bundle.parameters.database-name }}"
mysqlUser: "{{ bundle.parameters.mysql-user }}"
outputs:
- name: mysql-root-password
secret: porter-ci-mysql
key: mysql-root-password
- name: mysql-password
secret: porter-ci-mysql
key: mysql-password
With this bundle definition, we can build a second bundle to install wordpress and declare a dependency on this bundle. The porter.yaml
for this might look something like:
name: wordpress
version: 0.1.0
tag: getporter/wordpress
mixins:
- helm
dependencies:
- name: mysql
tag: getporter/mysql:v0.1.3
parameters:
database_name: wordpress
mysql_user: wordpress
credentials:
- name: kubeconfig
path: /root/.kube/config
parameters:
- name: wordpress-name
type: string
default: porter-ci-wordpress
env: WORDPRESS_NAME
- name: wordpress-password
type: string
sensitive: true
applyTo:
- install
- upgrade
- name: namespace
type: string
default: ''
install:
- helm:
description: "Install Wordpress"
name: "{{ bundle.parameters.wordpress-name }}"
chart: stable/wordpress
namespace: "{{ bundle.parameters.namespace }}"
replace: true
set:
wordpressPassword: "{{ bundle.parameters.wordpress-password }}"
externalDatabase.password: "{{ bundle.dependencies.mysql.outputs.mysql-password }}"
externalDatabase.port: 3306
mariadb.enabled: false
outputs:
- name: wordpress-password
secret: "{{ bundle.parameters.wordpress-name }}"
key: wordpress-password
The wordpress bundle declares a dependency on the mysql
bundle, which we saw above. Now, we are able to refer to the parameters and the outputs from that bundle!
install:
- helm:
description: "Install Wordpress"
name: "{{ bundle.parameters.wordpress-name }}"
chart: stable/wordpress
namespace: "{{ bundle.parameters.namespace }}"
replace: true
set:
wordpressPassword: "{{ bundle.parameters.wordpress-password }}"
externalDatabase.password: "{{ bundle.dependencies.mysql.outputs.mysql-password }}"
externalDatabase.port: 3306
mariadb.enabled: false
For more information on how dependencies are handled, refer to the dependencies documentation.
Combining References
It is possible to reference multiple parameters, credentials and/or outputs in a single place. You can combine the expressions as follows:
install:
- helm:
description: "Install Java App"
name: "{{ bundle.parameters.cool-app}}"
chart: stable/wordpress
replace: true
set:
jdbc_url: "jdbc:mysql://{{ bundle.outputs.mysql_host }}:{{ bundle.outputs.mysql_port }}/{{ bundle.parameters.database_name }}"