Adoption SIG: Meeting notes for Feb 16, 2023

[taras]

Participants

• Taras Mankovski
• Serena Nichols
• Francesco Corti
• Tom Coufal
• Waldir Montoya
• Phil K
• Min Kim

Update from Backstage for Backstage Working Group

@taras shared that Backstage for Backstage WG met to discuss the approach. Meeting notes can be found in this document. The group decided to focus on functionality that doesn't have an overlap with the microsite.

The first two features to be implemented are

• Show a list of all Backstage plugins published to NPM
• Allow using the scaffolder to create a new plugin

Update from Backstate Microservice Plugins Working Group

@taras shared that Backstage Microservice Plugins Working Group had their first meeting. @taras created a meta issue #68 to track progress of the group. Red Hat folks brought lots of experience from their work with Webpack Federation to create a pluggable framework for OpenShift's console. The group agreed to think about the discussion and reconvene in the near future. The next meeting is set for Feb 28th. To track the group's progress, subscribe to notifications of #68

Discord Analytics

@waldirmontoya25 shared that he's been going through Discord looking for all of the discussions where @freben and @Rugvip talked about extending the data model. It's time consuming and he wanted to know if there is a way to get analytics from Discord. It would be helpful to know what people are asking about. @fcorti said that he'll enquire internally and get back to the group.

Reading Secrets Asynchronously

@RobotSail shared that he and @taras were talking about contributions to Backstage that could be relevant for Red Hat. @taras shared a link to backstage/backstage#14084 and backstage/backstage#13952 which talk about the need for loading secrets without using environment variables or without storying them in memory at all. This is important for financial institutions that want to adopt Backstage.

In summary, there are 3 approaches that could allow using secrets in Backstage without using environment variables.

1. Allow injecting secrets into config in packages/backend/src/index.ts as described in Make it easier to merge configuration into config at runtime backstage#14084
2. Introduce an integration with vaults as described in Add native support for secret managers to Backstage config backstage#13952
3. Introduce a secret clients API that would allow fetching clients on demand

Option 1 is most approachable so we're going to start with that approach. Discussion is here

Opinionated Catalog Ingestion

@fcorti shared that he would like to restart the conversation about Opinionated Catalog Ingestion. He said that he doesn't have a solution in mind, but the wants to start discussing the problems that we'd want to solve with a more Opinionated Catalog Ingestion.

@fcorti shared that in previous discussions there was a strong message that requiring YAML to ingest into the catalog makes adoption more difficult. We'd like to find a way to populate the catalog when the organization doesn't have YAML documents or their documents are not in Backstage YAML format.

• @waldirmontoya25 shared that VMWare would like to support this effort.
• @adobejmong shared that Adobe is interested in this initiative as well.
• Greg Schultz shared that at their company they have lots of Spring Cloud Config files. It would useful for them to be able to ingest those instead of expecting catalog-info.yaml files to exist
• Ravi Tadikonda of USBank would like to engage in a conversation about using a push based model of ingestion with events instead of pull based.

Action item: @fcorti will setup a dedicated meeting and invite anyone interested in this topic. If you'd like to be added to the list, email francescoc@spotify.com

Demo of Embedded Scaffolder Workflow

Watch the demo video https://www.youtube.com/watch?v=vdpm4m7uMgg

@taras shared that the alpha version of Backstage Scaffolder Plugin includes a new feature Frontside created for one of their clients. It's called the Embedded Scaffolder Workflow, making it easier to onboard components in the Catalog to use features that require updating annotations. In the demo video, @taras shows how an entity in the Catalog not using TechDocs can be onboarded using the Embedded Scaffolder Workflow. The MissingAnnotation component, which users find confusing, is replaced by a scaffolder workflow that creates a pull request with everything the entity needs to start using TechDocs. The pull request includes configuring the continuous integration system to build and upload docs to the storage container, updating the catalog-info.yaml file with necessary annotation, adds the docs folder and mkdocs.yaml file. Instead of giving the user work to do, it gives them a pull request that they can merge and show docs in Backstage.

You can find the example code here thefrontside/playhouse#224

Stayed tuned for an alpha of this release.

Spontanious discussions

What is the status of the Backend System?

According to Backstage System Alpha blog post, Backstage System is not yet ready for production. There isn't clear timeline yet of when it'll be in production. There is a meta issue backstage/backstage#13084 to track necessary changes to plugins