We're seeing a 100% failure rate on the grenade multinode jobs on 2 cloud providers. We suspect it's because tempest is trying to ssh on the private network which it doesn't have access to. Lets revert to the previous behaviour and see if it fixes it
This reverts commit 46a272b31a.
Change-Id: I50a6de8db7d74b2250a8ad166d13c39af37e1c6c