This may not fix all the problems in the tripleo jobs but adds in the
base set of things we've learned about making multinode jobs work in the
migrated zuulv3 legacy jobs.
First we only run the main run playbook against the primary node.
Otherwise it runs on all the nodes and gets confused.
Second we open up the firewall between all the nodes so that they can
talk to each other.
Change-Id: Ibcba1bd338eeb7e9ccf983085838ef2a65ab647f
This should be a no-op for single-node jobs, but for multi-node jobs it
should keep them from trying to copy logs from all of their nodes,
which they are not designed to do.
Change-Id: Ibcf158f1b8082fbffeb48fa48c6592c87e056d01
This isn't really a thing in v3, but structurally it's a thing we
need to provide to v2 jobs so they don't get confused.
Change-Id: Icb528444f44f18ca29e02154513b9facefd04c89
This is the result of having run the zuul migration tool.
Depends-On: I72687a56285a3d733a9adeaf3dc037a1ac95efd3
Change-Id: Ib511da75908e152e72fdf0d7b496f4fa98f9c223
Signed-off-by: Paul Belanger <pabelanger@redhat.com>