Description of problem:
When using "oc replace" on routes, this causes the router to be unable to properly reload the routes until you restart the router itself.

Version-Release number of selected component (if applicable):
I replicated the issue in     Image:	openshift3/ose-haproxy-router:v3.2.1.13

Customer reported issue in

OpenShift Enterprise v3.2.1.15 (containerized)

HAProxy image is: openshift3/ose-haproxy-router:v3.2.1.15



How reproducible:
Reproduced

Steps to Reproduce:
1. oc create route
2. oc replace route
3. oc delete route
4. oc create route
5. oc get route

There's also a script in the github link (in additional info) that replicates the issue more easily. Just run the script with an existing service name as an argument then run oc get route.



Actual results:

$ ./routerscript jenkins-1-centos7
route "testrute" created
route "testrute" replaced
route "testrute" deleted
route "testrute" created
route "testrute" replaced
route "testrute" deleted
route "testrute" created
$ oc get route
NAME       HOST/PORT     PATH      SERVICE                  TERMINATION   LABELS
testrute   NoHostValue             jenkins-1-centos7:http                 


Expected results:

$ ./routerscript jenkins-1-centos7
route "testrute" created
route "testrute" replaced
route "testrute" deleted
route "testrute" created
route "testrute" replaced
route "testrute" deleted
route "testrute" created
$ oc get route
NAME       HOST/PORT     PATH      SERVICE                  TERMINATION   LABELS
testrute   <some hostname>         jenkins-1-centos7:http                 


Additional info:
https://github.com/openshift/origin/issues/8388