search

ovn-org / ovn-kubernetes

A robust Kubernetes networking platform
https://ovn-kubernetes.io/
Apache License 2.0
807 stars 338 forks source link

Test flakes in CI: Network Granular Checks: services #1293

Open trozet opened 4 years ago

trozet commented 4 years ago

We keep seeing these fail randomly in CI:

Summarizing 4 Failures:

[Fail] [sig-network] Networking Granular Checks: Services [It] should function for node-Service: udp /home/runner/go/src/k8s.io/kubernetes/_output/local/go/src/k8s.io/kubernetes/test/e2e/framework/network/utils.go:381

[Fail] [sig-network] Networking Granular Checks: Services [It] should function for endpoint-Service: udp /home/runner/go/src/k8s.io/kubernetes/_output/local/go/src/k8s.io/kubernetes/test/e2e/framework/network/utils.go:269

[Fail] [sig-network] Networking Granular Checks: Services [It] should update nodePort: udp [Slow] /home/runner/go/src/k8s.io/kubernetes/_output/local/go/src/k8s.io/kubernetes/test/e2e/framework/network/utils.go:381

[Fail] [sig-network] Networking Granular Checks: Services [It] should function for pod-Service: udp /home/runner/go/src/k8s.io/kubernetes/_output/local/go/src/k8s.io/kubernetes/test/e2e/framework/network/utils.go:269

Example: https://github.com/ovn-org/ovn-kubernetes/pull/1288/checks?check_run_id=610671712

danwinship commented 4 years ago

Skimming those results quickly, it is the NodeIP checks that are failing, not ClusterIP, and they appear to be failing completely (the NodeIP does not reach any of the three Endpoints), meaning this has nothing to do with the very similar set of flakes in OpenShift with openshift-sdn (which is good because we were very certain that those flakes were VXLAN-related and wouldn't affect OVN).