Successfully healthcheck only after generating the right template

[gianrubio]

1- Today I migrated the nginx controller to the latest version, using deployment strategy. During the deployment my ingress was down because the template was changed. After I adapt the template to the new version, the ingress was back.

The log was

[nginx-ingress-controller-54543-yvxty] W0110 10:14:03.384377       7 queue.go:87] requeuing kube-system/mypod, err template: nginx.tmpl:498:38: executing "nginx.tmpl" at <.PassthrougBackends>: can't evaluate field PassthrougBackends in type config.TemplateConfig
[nginx-ingress-controller-54543-yvxty] I0110 10:14:03.418421       7 leaderelection.go:247] lock is held by nginx-ingress-controller-54543-319ue and has not yet expired

For few seconds my applications was down. To avoid other users having the same issue, I propose to change the behaviour of healthcheck to only accept connections after generating the full template.

2- I'm not sure if when I was migrating the controller (with the right template) I got some nginx errors, probably because the nginx was generating the new template and using the default config, so my upstreams are empty. Can anyone confirm this issues?

[aledbf]

@gianrubio please do not use the current master. If you want to test the controller please apply the PR #95

[gianrubio]

@aledbf got it, could you confirm if this issues will be solved int this PR? Can I close this issue?

[aledbf]

@gianrubio yes, is fixed

[aledbf]

@gianrubio I want to merge #95 after another review.