Skip to content

CI: K8sServiceTest Tests NodePort BPF #8945

@brb

Description

@brb

The test suite seems to be flaky. The example failure:

/home/jenkins/workspace/Cilium-PR-Ginkgo-Tests-Validated/k8s-1.10-gopath/src/github.com/cilium/cilium/test/ginkgo-ext/scopes.go:420
Pod "testclient-6zb5g" can not connect to service "http://127.0.0.1:32174"
Expected command: kubectl exec -n default testclient-6zb5g -- curl --path-as-is -s -D /dev/stderr --fail --connect-timeout 5 --max-time 8 http://127.0.0.1:32174 -w "time-> DNS: '%{time_namelookup}(%{remote_ip})', Connect: '%{time_connect}',Transfer '%{time_starttransfer}', total '%{time_total}'" 
To succeed, but it failed:
Exitcode: 28 
Stdout:
 	 time-> DNS: '0.000021()', Connect: '0.000000',Transfer '0.000000', total '5.000594'
Stderr:
 	 command terminated with exit code 28 

The following instances of the failure were observed:

  • From pod curl to ClusterIP.
  • From pod curl to 127.0.0.1:NodePort
  • From pod curl to PublicIP:NodePort

The tunnel / direct routing mode setting does not matter, as the test failed in both modes. In all cases bpf_sock.c (aka host-reachable services) was involved.

/cc @borkmann

Metadata

Metadata

Assignees

Labels

area/CIContinuous Integration testing issue or flakeci/flakeThis is a known failure that occurs in the tree. Please investigate me!kind/bugThis is a bug in the Cilium logic.priority/highThis is considered vital to an upcoming release.

Type

No type

Projects

No projects

Milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions