incomprehensible error message

[symbiont-ji]

In

kustomize/api/loader/fileloader.go

Line 161 in a5914ab

"got file '%s', but '%s' must be a directory to be a root",

, what does "must be a directory to be a root" even mean? That's not English!

[Shell32-Natsu]

This error message means the path must be a directory, so it can be used as a build root.

[symbiont-ji]

Can the phrasing be fixed please?

…

On Fri, Apr 16, 2021 at 1:36 PM Donny Xia ***@***.***> wrote: This error message means the path must be a directory, so it can be used as a build *root*. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#3812 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ARFTEDW4674JDWZEVSVAQYLTJBYTDANCNFSM43AMZATA> .

[rdxmb]

This is indeed very confusing. I have a syntax error in my services.yaml , e.g. the ports key is missing. As a result I get

$ kustomize build .
Error: accumulating resources: accumulation err='accumulating resources from 'resources/service.yaml': yaml: line 9: did not find expected key': got file 'service.yaml', but '/home/marc/projects/k8-orchestration/tick-kustomize/base/kapacitor/resources/service.yaml' must be a directory to be a root

The first part of the error is kind of ok, but the second part does not make any sense here for the user.

Files:

$ cat resources/service.yaml 
---
apiVersion: v1
kind: Service
metadata:
  name: kapacitor
  labels:
    app.kubernetes.io/name: tick-kapacitor
spec:
  selector:
    app.kubernetes.io/name: tick-kapacitor
    - name: http
      port: 9092
      protocol: TCP
  type: ClusterIP

$ cat kustomization.yaml 
---
apiVersion: kustomize.config.k8s.io/v1beta1
kind: Kustomization
resources:
  - resources/service.yaml
  - resources/statefulset.yaml

[rdxmb]

@Shell32-Natsu in some cases the error message might be correct. However, it appears also when there are some syntax problems within the files. Then it's very confusing. This seems to be an issue for me :)

[k8s-triage-robot]

The Kubernetes project currently lacks enough contributors to adequately respond to all issues and PRs.

This bot triages issues and PRs according to the following rules:

• After 90d of inactivity, lifecycle/stale is applied
• After 30d of inactivity since lifecycle/stale was applied, lifecycle/rotten is applied
• After 30d of inactivity since lifecycle/rotten was applied, the issue is closed

You can:

• Mark this issue or PR as fresh with /remove-lifecycle stale
• Mark this issue or PR as rotten with /lifecycle rotten
• Close this issue or PR with /close
• Offer to help out with Issue Triage

Please send feedback to sig-contributor-experience at kubernetes/community.

/lifecycle stale

[m-Bilal]

/assign

[k8s-triage-robot]

The Kubernetes project currently lacks enough active contributors to adequately respond to all issues and PRs.

This bot triages issues and PRs according to the following rules:

• After 90d of inactivity, lifecycle/stale is applied
• After 30d of inactivity since lifecycle/stale was applied, lifecycle/rotten is applied
• After 30d of inactivity since lifecycle/rotten was applied, the issue is closed

You can:

• Mark this issue or PR as fresh with /remove-lifecycle rotten
• Close this issue or PR with /close
• Offer to help out with Issue Triage

Please send feedback to sig-contributor-experience at kubernetes/community.

/lifecycle rotten

[m-Bilal]

/remove-lifecycle rotten

[rdxmb]

@Shell32-Natsu

added the triage/not-an-issue label on Apr 16

Why?

[digihunch]

In my case, the actual error was duplicate resource name.
This error message is very misleading and it appears before the actual error message.

[m-Bilal]

The reason why "directory must be a root" error message got appended to the actual error in @rdxmb's case is that we're only checking for HTTP error and proceeding if there's any other error here:

kustomize/api/internal/target/kusttarget.go

Lines 407 to 412 in c65ef48

	if errF := kt.accumulateFile(ra, path); errF != nil {
	// not much we can do if the error is an HTTP error so we bail out
	if errors.Is(errF, load.ErrorHTTP) {
	return nil, errF
	}
	ldr, err := kt.ldr.New(path)

Is there a specific reason we're ignoring other errors here?

[m-Bilal]

Does the error message "path must be a directory so that it can be used as a build root" make more sense?

[natasha41575]

Is there a specific reason we're ignoring other errors here?

What we are trying to do is first try to load the resource as a file. If there is an error, proceed to try to load the resource as a directory.

Perhaps it doesn't make the most sense and there may be a way to restructure the code to provide more specific error messages.

[m-Bilal]

I think the fix would be to make Read() in kyaml/kio/byteio_reader.go return a specific error type (like MalformedYAMLError / InvalidInputError) so that we can check for it in kusttarget.go's accumulateFile() and return if that is the error type. Is that the correct approach?

Another way I can think of doing this is to check if the path is not a directory and there's an error in kusttarget's accumulateFile(), return the error instead of trying to load the resource as a directory.

[felipetssilva]

In

kustomize/api/loader/fileloader.go

Line 161 in a5914ab

"got file '%s', but '%s' must be a directory to be a root",

, what does "must be a directory to be a root" even mean? That's not English!

I am leaving this for future reference:

This error is because we must pass a directory to the kubectl apply -k <kustomization_directory> (or kubectl kustomize <kustomization_directory>) command. For example : this 'k apply -k application/' works! and this k apply -k application/application/kustomize.yaml DOES NOT!!

Error message is confusing but the problem is simple :)

[gocpplua]

minio
├── argo-server-deployment.yaml
├── argo-server-sa.yaml
├── argo-server-service.yaml
└── kustomization.yaml

/data/kubernetes/k3s$ sudo k3s kubectl apply -k ./minio
serviceaccount/argo-server created
service/argo-server created
deployment.apps/argo-server created

[lugipfupf]

In case this might help someone else:
Running on a rather old 2.5.5 ArgoCD, we got the same error message. Reason was a kustomization.yaml with quite a lot of yaml references hiding a duplicate entry.