k8s e2e tests are failing because of missing typha

[OlegLoewen]

What happened:
While running tests on shoot clusters without typha, we noticed multiple k8s e2e tests are repeatedly failing with error message:

Feb 17 08:40:04.346: Failed to find expected responses:
Tries 34
Command curl -g -q -s 'http://100.64.1.202:8080/dial?request=hostname&protocol=http&host=100.64.0.228&port=8080&tries=1'
retrieved map[]
expected map[netserver-0:{}]

Failing k8s e2e tests:

• [sig-api-machinery] Aggregator Should be able to support the 1.10 Sample API Server using the current Aggregator [Conformance]
• [sig-cli] Kubectl client Simple pod should handle in-cluster config
• [sig-network] DNS should provide /etc/hosts entries for the cluster [LinuxOnly] [Conformance]
• [sig-network] DNS should provide DNS for pods for Hostname [LinuxOnly] [Conformance]
• [sig-network] DNS should provide DNS for pods for Subdomain [Conformance]
• [sig-network] DNS should provide DNS for services [Conformance]
• [sig-network] DNS should provide DNS for the cluster [Conformance]
• [sig-network] DNS should resolve DNS of partial qualified names for services [LinuxOnly] [Conformance]
• [sig-network] DNS should resolve DNS of partial qualified names for the cluster [LinuxOnly]
• [sig-network] Networking Granular Checks: Pods should function for intra-pod communication: http [LinuxOnly] [NodeConformance] [Conformance]
• [sig-network] Networking Granular Checks: Pods should function for intra-pod communication: udp [LinuxOnly] [NodeConformance] [Conformance]
• [sig-network] Services should be able to switch session affinity for service with type clusterIP [LinuxOnly]

What you expected to happen:
Normally, it shouldn't matter for k8s e2e tests, if shoot cluster has typha enabled or disabled.

How to reproduce it (as minimally and precisely as possible):

1. Create gcp shoot cluster with typha disabled
2. Set KUBECONFIG to your cluster
3. Run docker run -ti -e --rm -v $KUBECONFIG:/mye2e/shoot.config golang:1.13 bash
4. Run within container:

export E2E_EXPORT_PATH=/tmp/export; export KUBECONFIG=/mye2e/shoot.config
go get github.com/gardener/test-infra; cd /go/src/github.com/gardener/test-infra
export GO111MODULE=on
go run -mod=vendor ./integration-tests/e2e -debug=true -k8sVersion=1.17.3 -cloudprovider=gcp -testcase="[sig-api-machinery] Aggregator Should be able to support the 1.10 Sample API Server using the current Aggregator [Conformance]" -testcase="[sig-cli] Kubectl client Simple pod should handle in-cluster config" -testcase="[sig-network] DNS should provide /etc/hosts entries for the cluster [LinuxOnly] [Conformance]" -testcase="[sig-network] DNS should provide DNS for pods for Hostname [LinuxOnly] [Conformance]" -testcase="[sig-network] DNS should provide DNS for pods for Subdomain [Conformance]" -testcase="[sig-network] DNS should provide DNS for services  [Conformance]" -testcase="[sig-network] DNS should provide DNS for the cluster  [Conformance]" -testcase="[sig-network] DNS should resolve DNS of partial qualified names for services [LinuxOnly] [Conformance]" -testcase="[sig-network] DNS should resolve DNS of partial qualified names for the cluster [LinuxOnly]" -testcase="[sig-network] Networking Granular Checks: Pods should function for intra-pod communication: http [LinuxOnly] [NodeConformance] [Conformance]" -testcase="[sig-network] Networking Granular Checks: Pods should function for intra-pod communication: udp [LinuxOnly] [NodeConformance] [Conformance]" -testcase="[sig-network] Services should be able to switch session affinity for service with type clusterIP [LinuxOnly]"

Environment:

• Gardener version (if relevant): v1.0.4
• Extension version: v1.3.0
• Kubernetes version (use kubectl version): v1.17.3
• Cloud provider or hardware configuration: gcp

[zanetworker]

@ialidzhikov is this still relevant?

[ialidzhikov]

To my knowledge there was no PR to address this. So it should it be.