From 8a265f1696916d8334de7a7a4e56628f82f5f5b2 Mon Sep 17 00:00:00 2001 From: ronaldngounou Date: Tue, 11 Aug 2026 19:30:20 +0000 Subject: [PATCH] Wait 120s after cluster is ACTIVE before connecting in awscli-cp-with-pcp The loadtest can fail during connect with a DNS 'no such host' error for the cluster API server endpoint, because the endpoint instances take time to come up and the Route53 record has a ~60s TTL. Wait after the cluster is ACTIVE (create succeeded) before kubectl connects. --- .../tekton-resources/tasks/setup/eks/awscli-cp-with-pcp.yaml | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/tests/tekton-resources/tasks/setup/eks/awscli-cp-with-pcp.yaml b/tests/tekton-resources/tasks/setup/eks/awscli-cp-with-pcp.yaml index 44e88b25..2e1e6b8e 100644 --- a/tests/tekton-resources/tasks/setup/eks/awscli-cp-with-pcp.yaml +++ b/tests/tekton-resources/tasks/setup/eks/awscli-cp-with-pcp.yaml @@ -61,6 +61,11 @@ spec: aws eks create-cluster --name $(params.cluster-name) --region $(params.region) --kubernetes-version $(params.kubernetes-version) --role-arn $SERVICE_ROLE_ARN --resources-vpc-config subnetIds=$subnets,securityGroupIds=$sg --control-plane-scaling-config tier=$(params.control-plane-tier) $ENDPOINT_FLAG fi aws eks $ENDPOINT_FLAG --region $(params.region) wait cluster-active --name $(params.cluster-name) + # Cluster is ACTIVE (create succeeded). Wait for the API server endpoint to + # become resolvable/discoverable before connecting with kubectl: the endpoint + # instances take time to come up and the Route53 record has a ~60s TTL, so + # connecting immediately can fail with "no such host". + sleep 120 - name: write-kubeconfig image: alpine/k8s:1.35.0 script: |