Description of problem: CREATE: 2022-07-11T06:03:47.665492352Z I0711 06:03:47.665439 1 obj_retry.go:1364] Creating *v1.Pod test/hello-pod took: 727ns 2022-07-11T06:03:47.665768460Z I0711 06:03:47.665747 1 obj_retry.go:1364] Creating *factory.egressIPPod test/hello-pod took: 139.716µs 2022-07-11T06:03:47.672577508Z I0711 06:03:47.672538 1 pods.go:369] [test/hello-pod] creating logical port for pod on switch ip-10-0-53-79.us-east-2.compute.internal 2022-07-11T06:03:47.672847676Z I0711 06:03:47.672767 1 kube.go:73] Setting annotations map[k8s.ovn.org/pod-networks:{"default":{"ip_addresses":["10.129.2.31/23"],"mac_address":"0a:58:0a:81:02:1f","gateway_ips":["10.129.2.1"],"ip_address":"10.129.2.31/23","gateway_ip":"10.129.2.1"}}] on pod test/hello-pod 2022-07-11T06:03:47.719710933Z I0711 06:03:47.719672 1 pods.go:358] [test/hello-pod] addLogicalPort took 47.13852ms, libovsdb time 4.248335ms, annotation time: 42.362315ms COMPLETED: 2022-07-11T06:08:19.254219737Z I0711 06:08:19.254168 1 obj_retry.go:1276] Detected object test/hello-pod of type *factory.egressIPPod in terminal state (e.g. completed) during update event: will remove it WE DIDN'T DELETE ANYTHING KAPI GONE! 2022-07-11T06:08:19.266628736Z W0711 06:08:19.266585 1 obj_retry.go:1401] Unable to get *v1.Pod test/hello-pod from informer cache (perhaps it was already deleted?), skipping update: pod "hello-pod" not found 2022-07-11T06:08:19.266740988Z W0711 06:08:19.266712 1 obj_retry.go:1401] Unable to get *v1.Pod test/hello-pod from informer cache (perhaps it was already deleted?), skipping update: pod "hello-pod" not found 2022-07-11T06:08:19.266740988Z W0711 06:08:19.266734 1 obj_retry.go:1401] Unable to get *factory.egressIPPod test/hello-pod from informer cache (perhaps it was already deleted?), skipping update: pod "hello-pod" not found IGNORE DELETE EVENT 2022-07-11T06:08:19.266773661Z I0711 06:08:19.266756 1 obj_retry.go:1503] Ignoring delete event for completed resource *v1.Pod test/hello-pod 2022-07-11T06:08:19.266773661Z I0711 06:08:19.266764 1 obj_retry.go:1503] Ignoring delete event for completed resource *factory.egressIPPod test/hello-pod RECREATE NEW POD - ENDS UP WITH SAME IP AND STALE PORT 2022-07-11T06:08:20.555542658Z I0711 06:08:20.555498 1 obj_retry.go:1364] Creating *v1.Pod test/hello-pod took: 741ns 2022-07-11T06:08:20.563196606Z I0711 06:08:20.563156 1 obj_retry.go:1364] Creating *factory.egressIPPod test/hello-pod took: 7.549346ms 2022-07-11T06:08:20.566882323Z I0711 06:08:20.566842 1 pods.go:369] [test/hello-pod] creating logical port for pod on switch ip-10-0-53-79.us-east-2.compute.internal 2022-07-11T06:08:20.567060407Z I0711 06:08:20.567038 1 kube.go:73] Setting annotations map[k8s.ovn.org/pod-networks:{"default":{"ip_addresses":["10.129.2.31/23"],"mac_address":"0a:58:0a:81:02:1f","gateway_ips":["10.129.2.1"],"ip_address":"10.129.2.31/23","gateway_ip":"10.129.2.1"}}] on pod test/hello-pod 2022-07-11T06:08:20.614135395Z I0711 06:08:20.614095 1 pods.go:358] [test/hello-pod] addLogicalPort took 47.25565ms, libovsdb time 3.232352ms, annotation time: 43.386391ms Technically should be new bug, but let's not cause another 4 backport PRs and just fix it in this one itself. Opening a new bug to unblock verifying https://bugzilla.redhat.com/show_bug.cgi?id=2097243#c17 Version-Release number of selected component (if applicable): How reproducible: Steps to Reproduce: 1. 2. 3. Actual results: Expected results: Additional info: