Using Scoped Variables in Devtron

Introduction

In Devtron, many configuration values such as a database name, memory limit, or service endpoint may need to be used in multiple places. Instead of entering the same value repeatedly, you can store it in a scoped variable.

A scoped variable (key-value pair) allows you to define a value once and reuse it. The value of the variable will depend on the scopes mentioned below:

Category	Description
Global	Variable value will be universally same throughout Devtron.
Cluster	Variable value might differ for each Kubernetes cluster.
Environment	Variable value might differ for each environment within a cluster, e.g., staging, dev, prod.
Application	Variable value might differ for each application.
Environment + Application	Variable value might differ for each application on a specific environment.

For example, you can create a variable named db_name to store the database name and set its value to dev-db for development environment, and for production environment you can set its value to prod-db, at the time of deployment, Devtron will automatically choose the correct value based on the environment.

Advantages of using scoped variables

• Reduces repeatability: Configuration management team can centrally maintain the static data.

• Simplifies bulk edits: All the places that use a variable get updated when you change the value of the variable.

• Keeps data secure: You can decide the exposure of a variable's value to prevent misuse or leakage of sensitive data.

---

How to Define a Scoped Variable

Who can Perform this Action?

Only super admins can define scoped variables.

In Devtron, you can define scoped variables by using a YAML template. It will contain a schema for defining the scoped variables.

Depending on your use case, you can do either one of the following:

Action	Description
Download the YAML Template and define variables from scratch	In case you are defining the Scoped Variables for the first time: Download the YAML template to your local system. Define your variables Upload it back to the Devtron
Edit the existing saved YAML configuration	If you wish to add, update or delete the existing scoped variables, you can edit the existing saved YAML configuration using any of the following methods: Edit using the in-built UI Editor. You can download the existing saved configuration as a YAML file in your local system and can modify it in your preferred local editor and then upload the file to implement changes.

Download the Template

1. From the left sidebar, go to Global Configurations → Scoped Variables

2. Click Download template.

   

   Figure 1: Downloading the Template
3. Open the downloaded template using any code editor (say VS Code).

Enter the Values

The YAML file contains key-value pairs that follow the below schema:

Field	Type	Description
apiVersion	string	The API version of the resource (comes pre-filled)
kind	string	The kind of resource (i.e. Variable, comes pre-filled)
spec	object	The complete specification object containing all the variables
spec.name	string	Unique name of the variable, e.g. DB_URL
spec.shortDescription	string	A short description of the variable (up to 120 characters)
spec.notes	string	Additional details about the variable (will not be shown on UI)
spec.isSensitive	boolean	Whether the variable value is confidential (will not be shown on UI if true)
spec.values	array	The complete values object containing all the variable values as per context

The spec.values array further contains the following elements:

Field	Type	Description
category	string	The context, e.g., Global, Cluster, Application, Env, ApplicationEnv
value	string	The value of the variable
selectors	object	A set of selectors that restrict the scope of the variable
selectors.attributeSelectors	object	A map of attribute selectors to values
selectors.attributeSelectors.<selector_key>	string	The key of the attribute selector, e.g., ApplicationName, EnvName, ClusterName
selectors.attributeSelectors.<selector_value>	string	The value of the attribute selector

Here's a truncated template containing the specification of two variables for your understanding:

apiVersion: devtron.ai/v1beta1
kind: Variable
spec:

# First example of a variable
  - name: DB_URL
    shortDescription: My application's customers are stored
    notes: The DB is a MySQL DB running version 7.0. The DB contains confidential
      information.
    isSensitive: true
    values:
      - category: Global
        value: mysql.example.com

# Second example of a variable
  - name: DB_Name
    shortDescription: My database name to recognize the DB
    notes: NA
    isSensitive: false
    values:
      - category: Global
        value: Devtron
      - category: ApplicationEnv
        value: app1-p
        selectors:
          attributeSelectors:
            ApplicationName: MyFirstApplication
            EnvName: prod

Upload the Template

1. Once you save the YAML file, go back to the screen where you downloaded the template.

2. Click Upload file to add, to upload your saved YAML file.

   

   Figure 2: Uploading the Template
3. The content of the file will be uploaded for you to review and edit. Click Review Changes.

   

   Figure 3: Reviewing the YAML file
4. You may check the changes between the last saved file and the current one before clicking Save.

   

   Figure 4: Saving the file
5. Click the Variable List tab to view the variables. Check the How to Use a Scoped Variable section to know more.

   

   Figure 5: List of Variables

---

Defining YAML values in scoped variables

In Devtron, Scoped variables usually store simple values like strings or numbers, but, in some cases, you may need to define more detailed configuration in a scoped variable, such as autoscaling configuration or resource configuration. You can achieve this by defining a YAML snippet as the value of a scoped variable.

Defining YAML snippets as the value of a scoped variable will help you to reuse YAML configuration across multiple applications or environments. Scoped variables with YAML snippets can be changed based on the context (category) such as Global, Cluster, Environment, Application, Environment+Application.

Here's a truncated template containing the specification of one scoped variable with a YAML snippet as its value defined for different contexts for your understanding:

apiVersion: devtron.ai/v1beta1
kind: Variable
spec:
  - notes: Resource Configuration
    shortDescription: Scoped variable with YAML snippet
    isSensitive: false
    name: resources # Defining name for the variable.
    values:
      - category: Application
        value:  # Insert the YAML configuration block with proper indentation
          limits:
            cpu: 100m
            memory: 100Mi
          requests:
            cpu: 100m
            memory: 100Mi
        selectors:
          attributeSelectors:
            ApplicationName: banking-preprod
      - category: Env # Defining the variable for an environment.
        value: 
            limits:
              cpu: 75m
              memory: 75Mi
            requests:
              cpu: 50m
              memory: 50Mi
        selectors:
          attributeSelectors:
            EnvName: devtron-demo # Specifying the environment
      - category: Env # Defining variable's values for another environment
        value: 
          limits:
            cpu: 200m
            memory: 200Mi
          requests:
            cpu: 100m
            memory: 100Mi
        selectors:
          attributeSelectors:
            EnvName: banking # Specifying another environment name.
      - category: Global # Defining Variable's Values for global context.
        value:
          limits:
            cpu: 50m
            memory: 50Mi
          requests:
            cpu: 50m
            memory: 50Mi

---

How to Edit an Existing Scoped Variable

Who can Perform this Action?

Only super admins can edit scoped variables

Only a super-admin can edit existing scoped variables.

Option 1: Directly edit using the UI

Figure 6: Editing from UI

Option 2: Reupload the updated YAML file

Figure 7: Re-uploading New File

Re-uploading the YAML file will replace the previous file, so any variable that existed in the previous file but not in the latest one will be lost

Option 3: Edit through 'Environments' tab

The Environments tab allows you to view and edit scoped variable values for individual environments.

1. Go to the Environments tab; you will see a list of all environments and how many scoped variables are defined for each of them.

Figure 8: Navigating to 'Environment' tab

1. Click the preferred environment name to view or edit.

2. You can edit the variables using the GUI or YAML mode.

Figure 9: Editing in GUI Mode

Figure 10: Editing in YAML Mode

Note

Any changes you made through this method will also update the saved YAML configuration.

---

How to Use a Scoped Variable

Who can Perform this Action?

Users need to have Admin permission or above (along with access to the environment and application) to enable to use a scoped variable.

Once a variable is defined, it can be used by your authorized users on Devtron. A scoped variable widget would appear only on the screens that support its usage.

Currently, the widget is shown only on the following screens in App Configuration:

• Workflow Editor → Edit build pipeline → Pre-build stage (tab)

• Workflow Editor → Edit build pipeline → Post-build stage (tab)

• Workflow Editor → Edit deployment pipeline → Post-Deployment stage (tab)

• Workflow Editor → Edit deployment pipeline → Post-Deployment stage (tab)

• Deployment Template

• ConfigMaps

• Secrets

Figure 11: Unexpanded Widget

To use a scoped variable, click on the floating widget; a list of variables will be visible.

Figure 12: Expanded Widget

Use the copy button to copy a relevant variable of your choice.

Figure 13: Copying a Variable

It would appear in the following format upon pasting it within an input field: @{{variable-name}}.

Figure 14: Pasting a Variable

In case you are using a scoped variable in deployment template, you need to encapsulate it in double quotes i.e., "@{{variable-name}}"

Figure 15: Using Scoped Variable in Deployment Template

Figure 16: Performing a Dry Run

Note: Ignore the red underline while using a scoped variable in the deployment template.

---

Order of Precedence

When multiple values are associated with a scoped variable, the precedence order is as follows, with the highest priority at the top:

1. Environment + App

2. App

3. Environment

4. Cluster

5. Global

Example

Figure 17: Variable key in Red, Variable value in Green

1. Environment + App: This is the most specific scope, and it will take precedence over all other scopes. For example, the value of DB name variable for the app1 application in the prod environment would be app1-p, even though there is a global DB name variable set to Devtron. If a variable value for this scope is not defined, the App scope will be checked.

2. App: This is the next most specific scope, and it will take precedence over the Environment, Cluster, and Global scopes. For example, the value of DB name variable for the app1 application would be project-tahiti, even though the value of DB name exists in lower scopes. If a variable value for this scope is not defined, the Environment scope will be checked.

3. Environment: This is the next most specific scope, and it will take precedence over the Cluster and Global scopes. For example, the value of DB name variable in the prod environment would be devtron-prod, even though the value of DB name exists in lower scopes. If a variable value for this scope is not defined, the Cluster scope will be checked.

4. Cluster: This is the next most specific scope, and it will take precedence over the Global scope. For example, the value of DB name variable in the gcp-gke cluster would be Devtron-gcp, even though there is a global DB name variable set to Devtron-gcp. If a variable value for this scope is not defined, the Global scope will be checked.

5. Global: This is the least specific scope, and it will only be used if no variable values are found in other higher scopes. The value of DB name variable would be Devtron.

---

List of Predefined Variables

There are some system variables that exist by default in Devtron that you can readily use if needed:

• DEVTRON_NAMESPACE: Provides name of the namespace

• DEVTRON_CLUSTER_NAME: Provides name of the cluster configured on Devtron

• DEVTRON_ENV_NAME: Provides name of the environment

• DEVTRON_IMAGE_TAG: Provides image tag associated with the container image

• DEVTRON_IMAGE: Provides full image path of the container image, e.g., gcr.io/k8s-minikube/kicbase:v0.0.39

• DEVTRON_APP_NAME: Provides name of the application on Devtron

Currently, these variables do not appear in the scoped variable widget, but you may use them.