On Wed, Oct 17, 2018 at 1:45 PM Sahina Bose <sabose@redhat.com> wrote:


On Tue, Oct 16, 2018 at 6:54 PM Gianluca Cecchi <gianluca.cecchi@gmail.com> wrote:
Hello,
I would like to cleanup a 3 hosts HCI install that didn't complete.
If I connect to cockpit of second host it seems like an initial config screen.
But if I connect to cockpit of the first host, that I have used in first attempt and I have rebooted in the mean time, it seems to find an engine config and doesn't show choice of install from scratch.
On gluster engine volume I have already done clean up.
Is there anything I can do on host1 to cleanup its partial config?

have you tried ovirt-hosted-engine-cleanup ?


Not yet. I will try next time, as I think I have to run again the cockpit based gluster hci deploy a few times on host1, because of what described here:

For the moment, a poor-man cleanup that seems to work has been to execute on host1:

cd /gluster_bricks/engine/engine
rm -rf <run_generated_uuid>
rm -rf  __DIRECT_IO_TEST__ 
rm -rf .shard

cd /etc/vdsm/
cp -p vdsm.conf.20181016131252 vdsm.conf
systemctl restart vdsmd

cd /etc/ovirt-hosted-engine
rm -f hosted-engine.conf 

cd /etc/libvirt/
rm -rf storage
rm -f qemu/HostedEngineLocal.xml

shutdown -r now 

and then run deploy again.

In my case the deploy had stopped after engine storage domain creation and at step of guestfish to inject network configuration into the to-be hosted-engine VM
Possibly ovirt-hosted-engine-cleanup covers more areas and scenarios

Thanks in the mean time
Gianluca