kubeovn / kube-ovn / 26556046023

27%

push

github

fix: report VEG workload network readiness (#6790)

VpcEgressGateway status previously trusted Deployment readiness and silently skipped workload pods whose Multus attachment network status was missing. In that case a gateway pod could be Running and the Deployment could be fully ready while the VEG CR still reported Ready/Completed with fewer internal/external IPs than requested.

Collect workload pod status as part of VEG reconciliation, require every expected replica to be Running/Ready with pod IPs and external attachment IPs, and set Ready=False with WorkloadNetworkNotReady when the Deployment is ready but the attachment status is incomplete. Also enqueue the VEG when its workload pod changes so network-status annotation updates are reflected promptly.

Add unit coverage for the status collector and an e2e case that removes the workload pod attachment network-status annotation and expects the VEG to report not ready.

Signed-off-by: zhangzujian <zhangzujian.7@gmail.com>

39 of 73 new or added lines in 1 file covered. (53.42%)

14668 of 57524 relevant lines covered (25.5%)

0.59 hits per line