Bug 1268318

Description of problem:

Following on from
https://bugzilla.redhat.com/show_bug.cgi?id=1249890

I was mentioned that this issue would affect clients to,
see comments #13, #18, #20, #31, #32

Version-Release number of selected component (if applicable):

6.1.2 plus hotfix from https://bugzilla.redhat.com/show_bug.cgi?id=1249890

How reproducible:

100%

Steps to Reproduce:
1. Have client connected to satellite. Uses port :5646
2. Simulate dropped connection. I used iptables to drop all :5646 
   traffic from the client system
3. Satellite / Capsule side will drop and clean up the connections
   with the hotfix in place, the client side does not.

Actual results:

Client systems still think their connection is open, when it has
been torn down.

Expected results:

Client systems deal with torn down connections to the sat / capsule
and reconnect


Additional info:

Client system: 192.168.160.186 (appear to Satellite at 10.33.1.59 due to NAT)
Satellite sys: 10.33.9.118

Initial State:
- Client:
# netstat -tanp | grep :5646
tcp        0   5956 192.168.160.186:52656   10.33.9.118:5646        ESTABLISHED 14862/qdrouterd     
- Satellite:
# netstat -tanp | grep :5646
...   
tcp        0      0 10.33.9.118:5646        10.33.1.59:52656        ESTABLISHED 18866/qdrouterd 

Add iptables rules:
- Satellite netstat changes to
tcp        0   4597 10.33.9.118:5646        10.33.1.59:52656        FIN_WAIT1   - 
before disappearing completely shortly aftwards.

Client system still has
tcp        0      0 192.168.160.186:53051   10.33.9.118:5646        ESTABLISHED 14862/qdrouterd