[library] Created instances aren't reachable by ip after hard reset primary controller

[Andrey Sledzinskiy]

{

    "build_id": "2014-07-17_22-02-59",
    "ostf_sha": "9863db951a6e159f4fa6e6861c8331e1af069cf8",
    "build_number": "329",
    "auth_required": true,
    "api": "1.0",
    "nailgun_sha": "0af26b4d29a7f540911466f2374e256e8c899e60",
    "production": "docker",
    "fuelmain_sha": "079bf8c7d80d1c957343ac217747fdce00f2f85a",
    "astute_sha": "d90cad0130da014eded5c21fa5f31054ce999dac",
    "feature_groups": [
        "mirantis"
    ],
    "release": "5.1",
    "fuellib_sha": "baf62a223eaa610a47a964ee11110ee60830e5ac"

}

Steps:
1. Create next cluster - CentOS, HA, Neutron GRE, 3 controller, 2 compute, 1 cinder node
2. Deploy cluster
3. After successful deployment destroy primary controller and start it
4. Open Health Check tab
5. Run 'Check network connectivity from instance via floating IP' test

Expected - test passed successfully
Actual - test failed with 'Instance is not reachable by IP'

I tried to create instance manually and vnc to it - default gateway is not pingable from instance
It seems that there are no neutron routers on controllers after destroy/start primary

netns on controller with started vip is:

ip netns list
haproxy
qdhcp-5d95df82-c6b9-40da-b33a-49bc9b0fb928

Logs are attached