# Interview2: Design a sonobuoy plugin to do something cool

### Goal

Extend crashd with CAPG Provider.

#### Rationale

crashd currently supports CAPA & CAPV providers. GCP has been around for a long time and crashd should support kubernetes clusters running on Google Cloud Compute Engine that were deployed by Cluster API for feature parity.

#### Overview

crashd already has scaffolding that pulls information about worker nodes from Cluster API. Cluster API for GCP is not quite ready for production given the low number of adopters and GKE being one of of the best managed kubernetes offering.

#### Implementation

We can make a copy of the `starlark/capa_provider.go` file and use that as the basis for `capg_provider.go`. Unlike CAPA, CAPG doesn't provision a bastion instance, so you'll need to manually provision a bastion. I have implemented this feature.

https://github.com/vmware-tanzu/crash-diagnostics/pull/243