Config as Code (opslevel.yml)

In addition to defining services through OpsLevel’s UI and GraphQL API, you can also define services by placing a YAML file named opslevel.yml in a service’s repo. Leveraging our Git Repository Integrations, OpsLevel will automatically synchronize the contents of the opslevel.yml file whenever it changes. This config-as-code mechanism allows developers to retain their existing Git-based development workflows to keep this data up-to-date.

In addition to services, opslevel.yml can also track ownership and metadata about repositories. The contents of opslevel.yml differ depending on if it’s tracking a service or a repository. The schema below highlights these differences.

🚧

YAML Specification

OpsLevel uses YAML 1.1, to ensure that values are treated as strings it is highly recommended to use quotes (") around values. This ensures that tag values such as false are not converted to Booleans

Supported file extension: .yml

Only the .yml file extension is supported. .yaml will not be detected.

Ensure the file is namedopslevel.yml to be auto-detected.

Aliases

opslevel.yml uses human readable identifiers to specify the owner, or tier of a service or repository, as well as the lifecycle stage of a service. For example, an opslevel.yml that specifies the team that owns a service looks like:

service:
  name: Order Processor
  owner: orders_team

The identifier orders_team is an alias. Aliases are automatically generated by OpsLevel for teams, tiers, and lifecycles.

Finding the alias for a team

Navigate to a Team page. There you can find the team alias.

Finding the alias for a tier

Navigate to the Account page. From there you will see a list of all your Service Tiers and their aliases. Currently, only users with the Admin role have access to the Account page. Non-Admin users can access this information with the OpsLevel CLI using the opslevel list tier command.

Finding the alias for a lifecycle

Similar to Service Tiers, navigate to the Account page. From there you will see a list of all your Service Lifecycles and their aliases. Currently, only users with the Admin role have access to the Account page. Non-Admin users can access this information with the OpsLevel CLI using the opslevel list lifecycle command.

Alias Stability

Aliases are stable identifiers. If you rename an entity that has an alias (e.g., a team), OpsLevel will generate a new alias. However, the old alias will still be valid so any existing references to it from other opslevel.yml files will continue to work.

Converting an existing service to use opslevel.yml

If you have a service already defined in OpsLevel and want it to be managed with opslevel.yml, first ensure the service is already connected to a repo. When a service is connected to a repo, OpsLevel will know how to merge a newly detected opslevel.yml with the service. After the repo is connected to the service, you can download the service’s opslevel.yml file and commit it to the repo.

Step 1: Visit a service’s Summary tab and select the repository your service lives in by clicking Edit and + Add Repository.

Step 2: From the service page click the Download opslevel.yml button.

Step 3: Copy the contents or Download the opslevel.yml and commit to the repo you selected in Step 1.

Converting an existing repository to use opslevel.yml

You can use opslevel.yml to manage properties directly on a repository, such as: owner, tier, and tags.

Step 1: Select a repository that was imported using one of our Git Repository Integrations.

Step 2: From the repository page click the Download opslevel.yml button.

Step 3: Copy the contents or Download the opslevel.ymland commit to this repo.

Schema

The following tables provide an in depth list of all the properties supported by opslevel.yml.

Note: opslevel.yml currently only supports one top-level field of either repository or service. In the event that both are specified, the service section will take precedence.

Service

Repository

Attributes

Tags

Tools

Repositories

Dependencies

Alert Sources

JSON Schema

You can automatically validate your opslevel.yml file as part of your CI using the following JSON schema: opslevel.schema.yml

You can run the following command which uses ajv-cli.

ajv test -s opslevel.schema.yml -d opslevel.yml --valid

Locking

When a service or repository is managed by opslevel.yml, certain fields are locked. This locking applies across all methods of modifying the service, whether in the UI or various methods that go through the GraphQL API. This ensures that there is never a conflict between what’s defined in opslevel.yml and what’s defined in OpsLevel.

For services the name, description, owner, tier, system, lifecycle stage, language, and framework fields are all locked for editing.

For repositories the owner and tier fields are locked for editing.

Service and repository properties that are lists, such as tools, tags, aliases, dependencies, and repositories, can have new entries added in the UI, though existing entries defined in opslevel.yml will be locked.

As shown in the image below, to tell whether a service or repository is managed by opslevel.yml, you can look for a lock icon on the respective information page.

Locked Service

Locked Repository

Locked Service Dependency

Service locked from being added to a System

Resyncing the opslevel.yml

If a config file error occurs due to a change elsewhere (ex. a service dependency has been renamed or alias has been deleted), the Resync button can be used to update the service with the latest fetched opslevel.yml after making the necessary change (ex. add the referenced alias to the correct service). This can be more convenient in some cases instead of unnecessary commits to trigger OpsLevel to re-evaluate the opslevel.yml.

Example service opslevel.yml

version: 1
service:
  name: Shopping Cart Service
  lifecycle: generally_available
  tier: tier_1
  system: web
  product: Retail Website
  owner: order_management_team
  language: Ruby
  framework: Rails
  description: Allows users to add/remove products in their virtual shopping carts
    prior to placing an order.
  aliases:
  - cart
  tags:
  - key: db
    value: mysql
  - key: kafka-topic
    value: cart-additions
  repositories:
  - name: DunderMifflin/shopping_cart
    path: "/"
    provider: github
    display_name: Cart Service Code
  - name: DunderMifflin/TerraformRepo
    path: "/production/shopping_cart"
    provider: bitbucket
    display_name: Terraform Config
  tools:
  - name: Confluence
    category: runbooks
    url: http://company.atlassian.com/confluence/runbooks
  - name: PagerDuty
    category: incidents
    url: https://your_account.pagerduty.com/services/PH99999
    environment: Production
  - name: Datadog
    category: metrics
    url: https://app.datadog.com/your_dashboard?env=prod
    environment: Production
  - name: Datadog
    category: metrics
    url: https://app.datadog.com/your_dashboard?env=staging
    environment: Staging
  dependencies:
    - alias: recommendation_service
      notes: Provides products recommendations to enhance the users shopping experience
    - alias: suggestions_service
  alert_sources:
    - type: pagerduty
      external_id: ABCD123
    - type: opsgenie
      external_id: 1234567-abcd-1234-abcdefgh1234
    - type: datadog
      external_id: 1234567

Example repository opslevel.yml

version: 1
repository:
  tier: tier_1
  owner: order_management_team
  tags:
  - key: version
    value: v1.0.0

Templating

If your organization uses template repositories for creating new services or repositories, you will benefit from using opslevel.defaults.yml as a templating mechanism.

opslevel.defaults.yml allows you to specify default values for your services or repositories that will be applied once you add an opslevel.yml file. Values set in the opslevel.defaults.yml file will be overwritten by values from the opslevel.yml file.

It is recommended you add an opslevel.defaults.yml file to your template repository with some basic defaults set; these are often in the form of a default language/framework, default tags, default tools, and possibly more.

Format

The format of opslevel.defaults.yml is identical to that of opslevel.yml. The only difference is that the opslevel.defaults.yml file limits which values you can set.

Values you can set from opslevel.defaults.yml:

Like with opslevel.yml, you can validate your opslevel.defaults.yml file using the following JSON schema: opslevel.defaults.schema.yml.

FAQs

Where do I put the opslevel.yml file?

Just add your opslevel.yml file(s) to your repo. If you specify a service definition in this file and this repo isn’t already mapped to a service in OpsLevel, a new service will be created.

Note: OpsLevel will only look for files that use the .yml extension.

Is opslevel.yaml supported?

No, at this time we do not support looking for opslevel.yaml. Please ensure the .yml extension is used.

For a mono-repo with many services, is it possible to define multiple opslevel.yml files in one repo or multiple services in a single file?

It is not possible to define multiple services in a single opslevel.yml file, however you can define multiple opslevel.yml files in different subdirectories of a repo. Add an opslevel.yml file into each subdirectory, and OpsLevel will automatically detect the yml file and create a new service in the catalog, or lock the existing service with the linked repo.

I have two services that are both locked by the same repo and opslevel.yml, how do I delete the duplicate service if it’s a duplicate?

If you attach a repo to a service that is already locking a different service, we will apply the opslevel.yml to the new service and subsequently lock it.

In order to delete this service, you must first delete the repository connection on the duplicate service, which will unlock the service and allow you to delete it.

How can I get a list of all opslevel.yml errors?

Our GraphQL API can return these errors on the configErrors field under Service and Repository objects.

If you have more questions?

Just send us an email at [email protected].