Skip to content

[not a flake] K8sServicesTest Checks ClusterIP Connectivity Checks service on same node #7902

@tgraf

Description

@tgraf

This is unlikely a CI flake but a real problem

Symptom:

/home/jenkins/workspace/Cilium-PR-Ginkgo-Tests-Validated/src/github.com/cilium/cilium/test/ginkgo-ext/scopes.go:410
cannot curl to service IP from host
Expected command: curl --path-as-is -s -D /dev/stderr --fail --connect-timeout 5 --max-time 8 http://10.108.198.211/ -w "time-> DNS: '%{time_namelookup}(%{remote_ip})', Connect: '%{time_connect}',Transfer '%{time_starttransfer}', total '%{time_total}'" 
To succeed, but it failed:
Exitcode: 7 
Stdout:
      time-> DNS: '0.000022()', Connect: '0.000000',Transfer '0.000000', total '3.205344'
Stderr:
      

/home/jenkins/workspace/Cilium-PR-Ginkgo-Tests-Validated/src/github.com/cilium/cilium/test/k8sT/Services.go:136

The services are plumbed correctly:

cmd: kubectl exec -n kube-system cilium-8dvs9 -- cilium service list
Exitcode: 0 
Stdout:
 	 ID   Frontend              Backend                   
	 1    10.104.198.251:2379   1 => 10.10.1.211:2379     
	                            2 => 10.10.0.188:2379     
	                            3 => 10.10.1.43:2379      
	 2    10.96.0.10:53         1 => 10.10.1.156:53       
	 3    10.96.0.1:443         1 => 192.168.36.11:6443   
	 6    10.108.198.211:80     1 => 10.10.0.233:80       
	                            2 => 10.10.0.44:80  

I've noticed that the demo app deployments do not define a readiness probe so maybe the pdo is not up and listening yet.

Metadata

Metadata

Assignees

Labels

kind/bugThis is a bug in the Cilium logic.priority/highThis is considered vital to an upcoming release.priority/insane 🚨http://cultofthepartyparrot.com/parrots/shipitparrot.gif

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions