Deleting Ingress controller pod causes all ingress statuses to be wiped

[egeland]

Setup:
Two Ingress (nginx) controller pods
Several Ingresses

Observed:
When deleting one of the two pods (the one that's holding the lock, according to the logs), all ingresses lose their loadbalancer entries, causing complete outages for everything associated.

Expected:
When deleting the lock-holder, the other pod should acquire the lock. No ingresses should lose loadbalancer entries.

Versions:

Client Version: version.Info{Major:"1", Minor:"6", GitVersion:"v1.6.0", GitCommit:"fff5156092b56e6bd60fff75aad4dc9de6b6ef37", GitTreeState:"clean", BuildDate:"2017-03-28T19:15:41Z", GoVersion:"go1.8", Compiler:"gc", Platform:"darwin/amd64"}
Server Version: version.Info{Major:"1", Minor:"5", GitVersion:"v1.5.4", GitCommit:"7243c69eb523aa4377bce883e7c0dd76b84709a1", GitTreeState:"clean", BuildDate:"2017-03-07T23:34:32Z", GoVersion:"go1.7.4", Compiler:"gc", Platform:"linux/amd64"}

The nginx-ingress used is gcr.io/google_containers/nginx-ingress-controller:0.9.0-beta.3

[aledbf]

@egeland please check if the image quay.io/aledbf/nginx-ingress-controller:0.86 solves this issue

Edit: the issue is related to the number of replicas that are running (more than one can be running in the same node)