Description of problem:
---------------------------------------
On adding a host to a gluster-enabled cluster from RHEVM, it does not create the management bridge 'rhevm' on the host if another bridge is already present. After reboot, the host goes to 'Non-operational' state with the following message in the events log -
Host <hostname> does not comply with the cluster <cluster-name> networks, the following networks are missing on host: 'rhevm'
Find engine logs attached.
Version-Release number of selected component (if applicable):
oVirt Engine Version: 3.2.0-5.el6ev
ovirt-host-deploy-1.0.0-0.1.beta.el6ev.noarch
How reproducible:
Always
Steps to Reproduce:
1. Add a host to a cluster managed by RHSC (this creates the bridge ovirtmgmt)
2. Now remove the host from RHSC, and try to add it to a gluster-cluster managed by RHEVM.
Actual results:
RHEVM does not create the management bridge 'rhevm' and gives the above described message in the logs.
Expected results:
Since RHEVM is detecting that ovirtmgmt is already a bridge, then it should be able to work with ovirtmgmt.
Additional info
It is not just 'another bridge', it is a management bridge:
2013-01-30 04:09:11 DEBUG otopi.plugins.ovirt_host_deploy.vdsm.bridge bridge._getInterfaceForDestination:337 interface for 10.70.37.115 is ovirtmgmt
2013-01-30 04:09:11 INFO otopi.plugins.ovirt_host_deploy.vdsm.bridge bridge._validation:725 Management channel interface ovirtmgmt is already a bridge
In this case we do not delete existing bridge, this applies to both the old vdsm-bootstrap and the new ovirt-host-deploy.
Yeah, there has been an old decision not to outsmart an admin who bothered to implement his management interface on top of a bridge. Bootstrapping is expect to work for clean-slate machines. If the host has a complex predefined network configuration, we do not want to erase it.