kubeshop/testkube

Docker image inspection error while using private registry

narsdk opened this issue · 4 comments

narsdk commented

Describe the bug
I use a private ECR registry storing all images - testkube related and custom containers executors.

Private registry is set using global values attached to chart:

global:
  imageRegistry: "***************.dkr.ecr.eu-central-1.amazonaws.com"

Testkube installation works as expected and tests are executed using images from correct registry but before pod with test starts I got error in API which block execution for 60 seconds until timeout in InspectDockerImage which, as it looks in error message, still uses default registry:

  "level": "error",
  "ts": 1710848355.6584322,
  "caller": "containerexecutor/tmpl.go:296",
  "msg": "Docker image inspection error",
  "error": "could not start process with command: skopeo, exited with code:1  error: exit status 1,output: time=\"2024-03-19T11:39:15Z\" level=fatal msg=\"Error parsing image name \\\"docker://tests-boiler-plate-qa-testkube-executor:0.1.1\\\": pinging container registry registry-1.docker.io: Get \\\"https://registry-1.docker.io/v2/\\\": dial tcp 54.198.86.24:443: i/o timeout\"\n",
  "stacktrace": "github.com/kubeshop/testkube/pkg/executor/containerexecutor.NewJobOptions,
    /home/runner/work/testkube/testkube/pkg/executor/containerexecutor/tmpl.go:296
    github.com/kubeshop/testkube/pkg/executor/containerexecutor.(*ContainerExecutor).createJob
    /home/runner/work/testkube/testkube/pkg/executor/containerexecutor/containerexecutor.go:281
    github.com/kubeshop/testkube/pkg/executor/containerexecutor.(*ContainerExecutor).Execute
    /home/runner/work/testkube/testkube/pkg/executor/containerexecutor/containerexecutor.go:239
    github.com/kubeshop/testkube/pkg/scheduler.(*Scheduler).startTestExecution
    /home/runner/work/testkube/testkube/pkg/scheduler/test_scheduler.go:124
    github.com/kubeshop/testkube/pkg/scheduler.(*Scheduler).executeTest
    /home/runner/work/testkube/testkube/pkg/scheduler/test_scheduler.go:101
    github.com/kubeshop/testkube/pkg/workerpool.Request[...].execute
    /home/runner/work/testkube/testkube/pkg/workerpool/service.go:43
    github.com/kubeshop/testkube/pkg/workerpool.worker[...]
    /home/runner/work/testkube/testkube/pkg/workerpool/service.go:108"
}

May that be a reason of some misconfiguration, any workaround or fix available?

Version / Cluster

  • Which testkube version? 1.16.25
  • What Kubernetes cluster? EKS
  • What Kubernetes version? v1.29.1-eks-508b6b3
vsukhin commented

hey, @narsdk yes, needs to be fixed

narsdk commented

Great :) Any chances that it will be cherry-picked for the next release?

vsukhin commented

Yes, we will do it next week