Description of problem:

While doing a new install today (3.5.2 on CentOS 7), deploy failed because the system couldn't resolve the host name for my storage server. In bug #1160423 dns broken during creating of the mgmt bridge, but in this case mgmt was on eth1.79 vlan, and DNS config was on eth0.1. The deploy basically overwrote resolv.conf with a blank file.

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

# rpm -qa|grep ovirt
ovirt-release35-003-1.noarch
ovirt-engine-sdk-python-3.5.2.1-1.el7.centos.noarch
ovirt-hosted-engine-setup-1.2.3-1.el7.centos.noarch
ovirt-hosted-engine-ha-1.2.5-1.el7.centos.noarch
libgovirt-0.3.1-3.el7.x86_64
ovirt-host-deploy-1.3.1-1.el7.noarch

How reproducible:


Steps to Reproduce:
1. configure DNS on one interface, ovirt on another
2. run --deploy
3. check resolv.conf

Actual results:
empty file

Expected results:
unchanged contents

Additional info:
workaround is to run 'service network restart', which restores resolv.conf, and then redeploy.