Kubernetes pod status doesn't reflect Fargate task status

[Jeffwan]

After I schedule two pods, pod status is changed to pending which is expected and after that, they were changed to ProviderFailed

➜ k get pods
NAME                                READY   STATUS    RESTARTS   AGE
nginx-deployment-568c9ff59d-hzj2z   0/1     Pending   0          3s
nginx-deployment-568c9ff59d-z5zdz   0/1     Pending   0          3s

➜ k get pods
NAME                                READY   STATUS           RESTARTS   AGE
nginx-deployment-568c9ff59d-hzj2z   0/1     ProviderFailed   0          7s
nginx-deployment-568c9ff59d-z5zdz   0/1     ProviderFailed   0          7s

The major reason is UpdatePod is not implemented.

aws-fargate/provider.go

Lines 129 to 132 in 3508966

	func (p *FargateProvider) UpdatePod(ctx context.Context, pod *corev1.Pod) error {
	log.Printf("Received UpdatePod request for %s/%s.\n", pod.Namespace, pod.Name)
	return errNotImplemented
	}

2021/01/24 16:28:27 Received GetPod request for default/nginx-deployment-568c9ff59d-hzj2z.
2021/01/24 16:28:27 Received GetPod request for default/nginx-deployment-568c9ff59d-z5zdz.
2021/01/24 16:28:27 Received UpdatePod request for default/nginx-deployment-568c9ff59d-hzj2z.
INFO[0074] Updated k8s pod status                        key=default/nginx-deployment-568c9ff59d-hzj2z method=createOrUpdatePod name=nginx-deployment-568c9ff59d-hzj2z namespace=default node=virtual-kubelet operatingSystem=Linux phase=Pending pod=nginx-deployment-568c9ff59d-hzj2z podPhase=Pending provider=aws reason=ProviderFailed uid=3c426899-01ed-41a0-85e0-f59071edb52f watchedNamespace= workerId=10
WARN[0074] requeuing [["default/nginx-deployment-568c9ff59d-hzj2z"]] due to failed sync  error="failed to sync pod \"default/nginx-deployment-568c9ff59d-hzj2z\" in the provider: not implemented by Fargate provider" key=default/nginx-deployment-568c9ff59d-hzj2z method=handleQueueItem node=virtual-kubelet operatingSystem=Linux provider=aws watchedNamespace= workerId=10

The status is not updated to Running even after Fargate task is ready.

Pod status needs to be fixed to reflect real Fargate task status.

[TomaszKlosinski]

Any updates on this?