I think I found a bug in a module of the ovirt-host-deploy script.
I'm (still) using F17 with the latest nightly being:
rpm -aq | grep ovirt
ovirt-engine-cli-3.2.0.7-1.20121217.git5eddf58.fc17.noarch
ovirt-engine-dbscripts-3.2.0-1.20121219.gitcd0d2cc.fc17.noarch
ovirt-iso-uploader-3.1.0-1.fc17.noarch
ovirt-engine-userportal-3.2.0-1.20121219.gitcd0d2cc.fc17.noarch
ovirt-engine-setup-3.2.0-1.20121219.gitcd0d2cc.fc17.noarch
ovirt-engine-tools-common-3.2.0-1.20121219.gitcd0d2cc.fc17.noarch
ovirt-engine-backend-3.2.0-1.20121219.gitcd0d2cc.fc17.noarch
ovirt-host-deploy-0.0.0-0.0.master.20121219.git817103d.fc17.noarch
ovirt-engine-sdk-3.2.0.5-1.20121217.gitc0ab704.fc17.noarch
ovirt-engine-restapi-3.2.0-1.20121219.gitcd0d2cc.fc17.noarch
ovirt-engine-genericapi-3.2.0-1.20121219.gitcd0d2cc.fc17.noarch
ovirt-engine-config-3.2.0-1.20121219.gitcd0d2cc.fc17.noarch
ovirt-engine-webadmin-portal-3.2.0-1.20121219.gitcd0d2cc.fc17.noarch
ovirt-host-deploy-java-0.0.0-0.0.master.20121219.git817103d.fc17.noarch
ovirt-engine-notification-service-3.2.0-1.20121219.gitcd0d2cc.fc17.noarch
ovirt-release-fedora-5-2.noarch
ovirt-image-uploader-3.1.0-1.fc17.noarch
ovirt-engine-3.2.0-1.20121219.gitcd0d2cc.fc17.noarch
ovirt-log-collector-3.1.0-1.fc17.noarch
I'm using a Cluster with the Gluster bit checked and when adding that host
it errors out with a rather annoying error: package 'v' not found.
Turns out that the last line of
Mgmt01/usr/share/ovirt-host-deploy/plugins/ovirt-host-deploy/gluster/packages.py
is wrong, there is a ',' missing. I changed it to:
def _packages(self):
self.packager.installUpdate(('vdsm-gluster',))
And when I now try to deploy my host in the gluster Cluster it tries to
install vdsm-gluster package and dependancies.
It now fails in /usr/share/vdsm/addNetwork where it complaines about
ConfigNetworkError: (21, 'Both PREFIX and NETMASK supplied')
This host has already been an ovirt installed host.
I could for testing remove all traces of ovirt but I have another host
where I tried that and now it stops its networking when vdsmd starts ;-(
Went ahead and removed PREFIX and reran deploy. The bridge ovirtmgmt is
correctly created BUT there is an error in the log:
INFO:root:Adding network ovirtmgmt with vlan=, bonding=, nics=['em1'],
bondingOptions=None, mtu=None, bridged=True, options={'blockingdhcp':
'true', 'UUID': 'e121a99a-994e-479d-8de1-a56c14315545',
'USERCTL': 'no',
'DNS2': '172.19.1.18', 'DNS1': '172.19.1.12',
'bootproto': 'none',
'onboot': 'yes', 'IPV6INIT': 'no'}
libvir: Network Driver error : Network not found: no network with matching
name 'vdsm-ovirtmgmt'
libvir: Network Driver error : Network not found: no network with matching
name 'vdsm-ovirtmgmt'
libvir: Network Driver error : Requested operation is not valid: cannot
set autostart for transient network
Looking at /etc/libvirt/qemu/networks/* shows indeed there are references
to vdsm-ovirtmgt.
Maybe this is because I'm also using fedora-virt-preview-repo because the
nightlies need a newer libvirt.
Let me know if a slightly lower version of libvirt will fix this.
Joop