Platform Engineering: Backstage & Crossplane: interview questions
The question you are most likely to get for each topic, a model answer, and what else comes up.
How would you design an internal developer platform for forty product teams? Mid
Start from the teams' cognitive load: shadow them, list what slows them down, and build the thinnest viable platform as a product. A portal (Backstage) with the software catalog, templates that create repositories and register them, and TechDocs; platform APIs (Crossplane XRDs and Compositions) that turn short requests into infrastructure with the company's defaults; GitOps to apply the requests; and golden paths that are supported but optional. Measure adoption and lead time, not just existence.
Also asked: What is a golden path and why should it be optional? · How does Crossplane differ from running Terraform in CI? · What would you put in the software catalog first?
What is an internal developer platform and why would a company build one? Mid
An IDP is a self-service layer a platform team runs as a product for stream-aligned teams. It usually combines a portal such as Backstage, platform APIs that turn requests into infrastructure such as Crossplane Compositions, and golden paths: supported, optional, opinionated ways to do common things. Companies build one to reduce cognitive load, so teams ship faster and the company's decisions on security, backups and ownership are applied once instead of by every team.
Also asked: What are the team types in Team Topologies? · What makes a golden path different from a mandatory process? · Which metrics show whether a platform helps?
Learn it: 42.1 Platform engineering: what an internal developer platform is
What is Crossplane and how does it work? Mid
Crossplane turns the Kubernetes API into a control plane for external infrastructure. Providers add managed resource kinds and run controllers that reconcile them against one external API, authenticated by a ProviderConfig. A platform team defines its own kinds with an XRD and says what to build for them in a Composition, a pipeline of functions; each request is a composite resource. It is installed with Helm into crossplane-system, and in v2 XRs are namespaced and claims are gone.
Also asked: What is the difference between a managed resource and a composite resource? · Why does the Crossplane Helm chart contain no CRDs? · What changed between Crossplane v1 and v2?
Learn it: 42.2 Crossplane v2: the Kubernetes API as a control plane
A Crossplane managed resource never gets any status. How do you debug it? Mid
No conditions at all means no controller is reconciling it, so check the provider first: kubectl get providers must show Installed and Healthy True, and its ProviderRevision and pod must be up. Then check the ProviderConfig the resource references - by default ClusterProviderConfig default - because a missing one gives a CannotConnectToProvider event. With InjectedIdentity, also check the provider ServiceAccount's RBAC, ideally fixed with a DeploymentRuntimeConfig.
Also asked: What is the difference between a Provider and a ProviderRevision? · What does a ProviderConfig do? · Why are AWS providers split into a family?
A managed resource shows SYNCED False and READY True. What does that mean? Mid
The external resource exists and was usable at the last successful observation, but the last reconcile failed - often credentials, permissions or an update the API refused. Read the Synced message from the right to find the cloud's error code, such as ExpiredToken. It is dangerous because applications keep working, so nothing alerts, while Crossplane no longer corrects drift or applies changes; alert on Synced False that lasts, and check the ClusterProviderConfig and its Secret.
Also asked: What is the difference between forProvider and atProvider? · How do you import an existing bucket into Crossplane? · What does managementPolicies Observe do?
Learn it: 42.6 Managed resources: Synced, Ready, external names and policies
How do you publish a self-service API with Crossplane? Mid
Write a CompositeResourceDefinition in apiextensions.crossplane.io/v2: a group, kind and plural, scope Namespaced, and a versioned OpenAPI schema with only the fields teams should choose, validated with enum and required. Crossplane generates the CRD and its controller. Then a Composition says what to build for that kind, and each request is a composite resource in the team's namespace; RBAC on the namespace decides who may ask.
Also asked: What fields does Crossplane add to every composite resource? · Why did Crossplane v2 remove claims? · How would you version a platform API?
How does a Crossplane Composition turn a request into resources? Mid
A Composition in Pipeline mode lists steps that call functions. With function-patch-and-transform, each resource has a name, a base manifest and patches such as FromCompositeFieldPath and ToCompositeFieldPath with transforms like map or string Format. Crossplane runs the pipeline, garbage collects resources that are no longer desired, applies the rest with server-side apply under stable names, and marks the XR Ready when every resource passes its readiness check. I test changes with crossplane composition render first.
Also asked: What does the map transform do with an unknown key? · Why would a composition fail with an RBAC error? · What is a PatchSet?
Learn it: 42.9 Compositions and function-patch-and-transform
A team's Crossplane XR is not Ready. Walk me through debugging it. Mid
I start with crossplane resource trace on the XR, -o wide for full messages, and stop at the first object that is not healthy. On the XR itself: composition selection (no compatible Compositions means compositeTypeRef does not match the apiVersion), the function, or a fatal function result. On a composed resource: connect failed means config or credentials, Unsynced means the API server refused it. Then provider Installed and Healthy, and crossplane composition render to reproduce the pipeline locally.
Also asked: What does a managed resource with no conditions tell you? · How would you find a typo in a patch? · How do you pause reconciliation?
How do you change a Crossplane Composition in production without losing data? Mid
Composed resources that leave the desired state are garbage collected, so a renamed or removed resource is deleted for every XR on Automatic updates. I protect stateful resources with managementPolicies without Delete, the cloud's deletion protection and a Usage; pin production XRs with compositionUpdatePolicy Manual; render the change with crossplane composition render and observed resources and diff it; let staging follow first; then move production XRs one at a time by setting their revision.
Also asked: What is a Crossplane Usage? · What is the difference between Automatic and Manual update policy? · What happens when you uninstall a provider before its resources?
What is the Backstage software catalog and how do services get into it? Mid
It holds entities - Component, API, Resource, System, Domain, Group, User - described by each team in a catalog-info.yaml in its own repository. You register the file's URL as a location; the catalog processes it on a loop, validates it against policies and the kind schemas, and derives relations such as ownedBy, partOf, providesApi and dependsOn. Plugins find external data through annotations. I test new files with a dry-run location or validate-entity before registering.
Also asked: What are the required fields of a Component? · What is a Location entity? · How would you keep ownership in the catalog accurate?
Learn it: 42.16 Backstage and the software catalog
How do Backstage Software Templates work? Mid
A Template is a catalog entity with parameters (a JSON Schema form, validated again on the backend), steps that run actions such as fetch:template, publish:gitlab or publish:github and catalog:register, templated with expressions, and output links. The scaffolder runs it as a task with a log. Steps are not transactional: a failure leaves the repository and the location behind, and catalog:register reports an invalid entity only as 400 Bad Request, so I dry-run the location to find the cause.
Also asked: What does fetch:template do? · How would you test a template in CI? · What happens if you run the same template twice?
Learn it: 42.18 Software Templates: the golden path as a form
How do TechDocs and the Kubernetes plugin find a component's docs and pods? Mid
TechDocs reads the backstage.io/techdocs-ref annotation, usually dir:., and builds the repository's Markdown with MkDocs and the techdocs-core plugin; in the recommended setup CI runs techdocs-cli generate and publish to object storage. The Kubernetes plugin queries each configured cluster for objects labelled backstage.io/kubernetes-id with the entity's value, per kind, so the Deployment and its pod template both need the label.
Also asked: What is the difference between the basic and recommended TechDocs setups? · Why keep documentation in the service repository? · How does the Kubernetes plugin authenticate to clusters?
Learn it: 42.20 TechDocs and the Kubernetes plugin
Walk me through a new service going to production on your platform. Mid
The developer runs a Software Template in Backstage; the scaffolder creates the repository with catalog-info.yaml, docs and a declarative request such as a Crossplane Database, publishes it and registers it in the catalog. GitOps applies the request to the team's namespace, and Crossplane builds the database from a reviewed Composition with the company's defaults. Backstage shows owner, relations, docs and the Kubernetes tab. The platform has its own SLOs and adoption metrics.
Also asked: What would you measure to prove the platform helps? · Where does the company's security policy live in this flow? · How do you roll out a change to the golden path?
Learn it: 42.31 Putting it together: one golden path, end to end
Practise these answers with flashcards and labs Free, in your browser - a real Ubuntu terminal to try it in, with missions that check your work.