I'm trying to set up a new ovirt cluster.  I got it "mostly working" earlier, but wanted to change some physical networking stuff, and so I thought I'd blow away my machines and rebuild.  I followed the same recipe to build it all, but now I'm failing at a point that previously worked.

I've built a 3 node cluster with glusterfs backing (3 brick replica), and all that is good and well.  I run the engine-setup --deploy, and it does its stuff, asks me (among other things) the admin password, I type in the password I want it to use (just like last time), then it says to log into the new VM and run engine-setup.  Here's the problem: I try to ssh in as root, and it will NOT accept my password.  It worked a couple days ago, doing it the exact same way, but it will not work now.

I've destroyed and re-deployed several times, I've even done a low level wipe of all three nodes and rebuild everything, and again, it doesn't work.

My only guess is that one of the packages the gdeploy script changed, and it has a bug or "new feature" that breaks this for some reason.  Unfortunately, I do not have the package versions that worked or the current list to compare to, so I cannot support this.

In any case, I'm completely stuck here...I can't log in to run engine-deploy, and I don't know enough of the console/low level stuff to try and hack my way into the VM (eg, to manually mount the disk image and replace the password or put my SSH key in).

Suggestions?  Can anyone else replicate this?

--Jim