Fatal deepcopy error on operator-sdk add api

[tomklino]

While following the getting started guide (created a memcached operator), and reaching the part of the guide that stated add a new custom-resource-definition, I issued the command operator-sdk add api --api-version=cache.example.com/v1alpha1 --kind=Memcached

Some files have changed but eventually the command failed like so:

memcached-operator (master) $ operator-sdk add api --api-version=cache.example.com/v1alpha1 --kind=Memcached
INFO[0000] Generating api version cache.example.com/v1alpha1 for kind Memcached. 
INFO[0000] Created pkg/apis/cache/group.go              
INFO[0000] Created pkg/apis/cache/v1alpha1/memcached_types.go 
INFO[0000] Created pkg/apis/addtoscheme_cache_v1alpha1.go 
INFO[0000] Created pkg/apis/cache/v1alpha1/register.go  
INFO[0000] Created pkg/apis/cache/v1alpha1/doc.go       
INFO[0000] Created deploy/crds/cache.example.com_v1alpha1_memcached_cr.yaml 
INFO[0000] Created deploy/crds/cache.example.com_memcacheds_crd.yaml 
INFO[0000] Running deepcopy code-generation for Custom Resource group versions: [cache:[v1alpha1], ] 
F1227 15:01:21.313323    2077 deepcopy.go:885] Hit an unsupported type invalid type for invalid type, from ./pkg/apis/cache/v1alpha1.Memcached

I'm not sure if I did something wrong but I followed the tutorial to the letter and tried recreating the project from scratch just in case, but ran into the same issue. Additionally, I could not find any similar case while googling the error

Versions:

$ operator-sdk version
operator-sdk version: "v0.13.0", commit: "1af9c95bb51420c55a7f7f2b7fabebda24451276", go version: "go1.13.3 linux/amd64"

$ go version
go version go1.13.5 linux/amd64

$ git version
git version 2.17.1

[tomklino]

Okay, so about 60 seconds after posting this, a very helpful person on the kubernetes slack pointed me to this issue.

tl;dr: the environment variable GOROOT should have been set like so

export GOROOT=$(go env GOROOT)

[estroz]

@tomklino I assume you were able to resolve your issue using the suggested workaround. If not, feel free to re-open this issue.