Error for .kube directory already existent with DeploymentAction

[kprakashcontender]

Hello Team,

I have a job configured with Deployment Action and the Kubernetes Integration on the project is configured with the Deployment File correctly. I am deploying a sample Flask Application, and the Pipeline Job Configuration has the below steps.

When the Workflow Runs, the following error is displayed.
[INFO] Starting step "DeployApplication"
[INFO] # Plugin Kubernetes Deployment Plugin v0.47.0+cds.13856 is ready
[INFO] Cannot create directory .kube : mkdir .kube: file exists
[INFO] # Details: Cannot create directory .kube : mkdir .kube: file exists
[INFO] # Status: Fail
[INFO] Stopping plugin...
[INFO] End of step "DeployApplication" (3s)

I have tried to locate the ".kube" and it is only found in three places as below.

I don't see a reason why this is a problem, as when I a manually apply the Deployment file using "kubectl apply -f it works absolutely fine. Any help on this would be very useful. Thanks.

[yesnault]

Hi @kprakashcontender ,
The plugin-kubernetes tries to create the .kube directory into the current root workspace: {{.cds.workspace}}. This workspace is delete after the end of the job.
Can you add a step script above the DeployApplication, with a ls -la command to check if the .kube is really here or not?
The {{.cds.workspace}}/.kube directory is deleted after the step DeployApplication.

[kprakashcontender]

Hi, @yesnault As you've suspected there is a ".kube" directory in the current workspace.

[yesnault]

ok, so the plugin needs :

• to delete the .kube directory
• OR mv it into another name during plugin execution
• OR no update -> display that the .kube exist as today

[kprakashcontender]

So, do you suspect that this could be a bug with the plugin. If yes, what would be your advice as a temporary fix?

[yesnault]

@kprakashcontender the temporary workaround is to delete .kube directory in the action script before.